diff options
Diffstat (limited to 'xlators/mgmt/glusterd/src')
-rw-r--r-- | xlators/mgmt/glusterd/src/Makefile.am | 2 | ||||
-rw-r--r-- | xlators/mgmt/glusterd/src/glusterd-handler.c | 103 | ||||
-rw-r--r-- | xlators/mgmt/glusterd/src/glusterd-hooks.c | 2 | ||||
-rw-r--r-- | xlators/mgmt/glusterd/src/glusterd-op-sm.c | 13 | ||||
-rw-r--r-- | xlators/mgmt/glusterd/src/glusterd-quota.c | 6 | ||||
-rw-r--r-- | xlators/mgmt/glusterd/src/glusterd-rpc-ops.c | 44 | ||||
-rw-r--r-- | xlators/mgmt/glusterd/src/glusterd-sm.c | 39 | ||||
-rw-r--r-- | xlators/mgmt/glusterd/src/glusterd-snapshot.c | 565 | ||||
-rw-r--r-- | xlators/mgmt/glusterd/src/glusterd-store.c | 498 | ||||
-rw-r--r-- | xlators/mgmt/glusterd/src/glusterd-store.h | 5 | ||||
-rw-r--r-- | xlators/mgmt/glusterd/src/glusterd-utils.c | 1648 | ||||
-rw-r--r-- | xlators/mgmt/glusterd/src/glusterd-utils.h | 54 | ||||
-rw-r--r-- | xlators/mgmt/glusterd/src/glusterd-volgen.c | 149 | ||||
-rw-r--r-- | xlators/mgmt/glusterd/src/glusterd-volgen.h | 3 | ||||
-rw-r--r-- | xlators/mgmt/glusterd/src/glusterd-volume-ops.c | 2 | ||||
-rw-r--r-- | xlators/mgmt/glusterd/src/glusterd-volume-set.c | 34 | ||||
-rw-r--r-- | xlators/mgmt/glusterd/src/glusterd.c | 20 | ||||
-rw-r--r-- | xlators/mgmt/glusterd/src/glusterd.h | 12 |
18 files changed, 2483 insertions, 716 deletions
diff --git a/xlators/mgmt/glusterd/src/Makefile.am b/xlators/mgmt/glusterd/src/Makefile.am index ac53e67ee..9b33edf4d 100644 --- a/xlators/mgmt/glusterd/src/Makefile.am +++ b/xlators/mgmt/glusterd/src/Makefile.am @@ -29,7 +29,7 @@ noinst_HEADERS = glusterd.h glusterd-utils.h glusterd-op-sm.h \ AM_CPPFLAGS = $(GF_CPPFLAGS) -I$(top_srcdir)/libglusterfs/src \ -I$(rpclibdir) -I$(CONTRIBDIR)/rbtree \ -I$(top_srcdir)/rpc/xdr/src -I$(top_srcdir)/rpc/rpc-lib/src \ - -I$(CONTRIBDIR)/uuid \ + -I$(CONTRIBDIR)/uuid -I$(CONTRIBDIR)/mount \ -DSBIN_DIR=\"$(sbindir)\" -DDATADIR=\"$(localstatedir)\" \ -DGSYNCD_PREFIX=\"$(libexecdir)/glusterfs\"\ -DSYNCDAEMON_COMPILE=$(SYNCDAEMON_COMPILE) $(XML_CPPFLAGS) diff --git a/xlators/mgmt/glusterd/src/glusterd-handler.c b/xlators/mgmt/glusterd/src/glusterd-handler.c index b111d2e54..e0373c774 100644 --- a/xlators/mgmt/glusterd/src/glusterd-handler.c +++ b/xlators/mgmt/glusterd/src/glusterd-handler.c @@ -3415,6 +3415,18 @@ set_probe_error_str (int op_ret, int op_errno, char *op_errstr, char *errstr, "in this state"); break; + case GF_PROBE_MISSED_SNAP_CONFLICT: + snprintf (errstr, len, "Failed to update " + "list of missed snapshots from " + "peer %s", hostname); + break; + + case GF_PROBE_SNAP_CONFLICT: + snprintf (errstr, len, "Conflict in comparing " + "list of snapshots from " + "peer %s", hostname); + break; + default: snprintf (errstr, len, "Probe returned with " "unknown errno %d", op_errno); @@ -3902,6 +3914,52 @@ glusterd_handle_cli_clearlocks_volume (rpcsvc_request_t *req) } static int +get_volinfo_from_brickid (char *brickid, glusterd_volinfo_t **volinfo) +{ + int ret = -1; + char *volid_str = NULL; + char *brick = NULL; + char *brickid_dup = NULL; + uuid_t volid = {0}; + xlator_t *this = NULL; + + this = THIS; + GF_ASSERT (this); + GF_ASSERT (brickid); + + brickid_dup = gf_strdup (brickid); + if (!brickid_dup) + goto out; + + volid_str = brickid_dup; + brick = strchr (brickid_dup, ':'); + if (!brick) { + gf_log (this->name, GF_LOG_ERROR, + "Invalid brickid"); + goto out; + } + + *brick = '\0'; + brick++; + uuid_parse (volid_str, volid); + ret = glusterd_volinfo_find_by_volume_id (volid, volinfo); + if (ret) { + /* Check if it is a snapshot volume */ + ret = glusterd_snap_volinfo_find_by_volume_id (volid, volinfo); + if (ret) { + gf_log (this->name, GF_LOG_WARNING, + "Failed to find volinfo"); + goto out; + } + } + + ret = 0; +out: + GF_FREE (brickid_dup); + return ret; +} + +static int get_brickinfo_from_brickid (char *brickid, glusterd_brickinfo_t **brickinfo) { glusterd_volinfo_t *volinfo = NULL; @@ -3944,13 +4002,14 @@ out: int __glusterd_brick_rpc_notify (struct rpc_clnt *rpc, void *mydata, - rpc_clnt_event_t event, void *data) + rpc_clnt_event_t event, void *data) { - xlator_t *this = NULL; - glusterd_conf_t *conf = NULL; - int ret = 0; - char *brickid = NULL; - glusterd_brickinfo_t *brickinfo = NULL; + char *brickid = NULL; + int ret = 0; + glusterd_conf_t *conf = NULL; + glusterd_brickinfo_t *brickinfo = NULL; + glusterd_volinfo_t *volinfo = NULL; + xlator_t *this = NULL; brickid = mydata; if (!brickid) @@ -3967,6 +4026,37 @@ __glusterd_brick_rpc_notify (struct rpc_clnt *rpc, void *mydata, switch (event) { case RPC_CLNT_CONNECT: + /* If a node on coming back up, already starts a brick + * before the handshake, and the notification comes after + * the handshake is done, then we need to check if this + * is a restored brick with a snapshot pending. If so, we + * need to stop the brick + */ + if (brickinfo->snap_status == -1) { + gf_log (this->name, GF_LOG_INFO, + "Snapshot is pending on %s:%s. " + "Hence not starting the brick", + brickinfo->hostname, + brickinfo->path); + ret = get_volinfo_from_brickid (brickid, &volinfo); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, + "Failed to get volinfo from " + "brickid(%s)", brickid); + goto out; + } + + ret = glusterd_brick_stop (volinfo, brickinfo, + _gf_false); + if (ret) { + gf_log (THIS->name, GF_LOG_ERROR, + "Unable to stop %s:%s", + brickinfo->hostname, brickinfo->path); + goto out; + } + + break; + } gf_log (this->name, GF_LOG_DEBUG, "Connected to %s:%s", brickinfo->hostname, brickinfo->path); glusterd_set_brick_status (brickinfo, GF_BRICK_STARTED); @@ -3992,6 +4082,7 @@ __glusterd_brick_rpc_notify (struct rpc_clnt *rpc, void *mydata, break; } +out: return ret; } diff --git a/xlators/mgmt/glusterd/src/glusterd-hooks.c b/xlators/mgmt/glusterd/src/glusterd-hooks.c index 352b6ba11..78730a564 100644 --- a/xlators/mgmt/glusterd/src/glusterd-hooks.c +++ b/xlators/mgmt/glusterd/src/glusterd-hooks.c @@ -181,7 +181,7 @@ glusterd_hooks_set_volume_args (dict_t *dict, runner_t *runner) goto out; runner_add_arg (runner, "-o"); - for (i = 1; (ret == 0); i++) { + for (i = 1; ret == 0; i++) { snprintf (query, sizeof (query), "key%d", i); ret = dict_get_str (dict, query, &key); if (ret) diff --git a/xlators/mgmt/glusterd/src/glusterd-op-sm.c b/xlators/mgmt/glusterd/src/glusterd-op-sm.c index fb5e097d9..9b130b4c6 100644 --- a/xlators/mgmt/glusterd/src/glusterd-op-sm.c +++ b/xlators/mgmt/glusterd/src/glusterd-op-sm.c @@ -1020,7 +1020,7 @@ glusterd_op_stage_reset_volume (dict_t *dict, char **op_errstr) { int ret = 0; char *volname = NULL; - gf_boolean_t exists = _gf_false; + int exists = 0; char msg[2048] = {0}; char *key = NULL; char *key_fixed = NULL; @@ -1068,6 +1068,7 @@ glusterd_op_stage_reset_volume (dict_t *dict, char **op_errstr) ret = -1; goto out; } + if (!exists) { ret = snprintf (msg, sizeof (msg), "Option %s does not exist", key); @@ -1819,7 +1820,7 @@ glusterd_op_set_volume (dict_t *dict) if (dict_count == 0) { ret = glusterd_volset_help (NULL, &op_errstr); if (ret) { - gf_log (this->name, GF_LOG_ERROR, "%s", + gf_log (this->name, GF_LOG_ERROR, "%s", (op_errstr)? op_errstr: "Volume set help internal error"); } @@ -2081,12 +2082,12 @@ glusterd_op_sync_volume (dict_t *dict, char **op_errstr, if (volname) { ret = glusterd_add_volume_to_dict (volinfo, rsp_dict, - 1); + 1, "volume"); vol_count = 1; } else { list_for_each_entry (volinfo, &priv->volumes, vol_list) { - ret = glusterd_add_volume_to_dict (volinfo, - rsp_dict, count); + ret = glusterd_add_volume_to_dict (volinfo, rsp_dict, + count, "volume"); if (ret) goto out; @@ -3262,7 +3263,7 @@ glusterd_is_get_op (xlator_t *this, glusterd_op_t op, dict_t *dict) if (op == GD_OP_STATUS_VOLUME) return _gf_true; - if ((op == GD_OP_SET_VOLUME)) { + if (op == GD_OP_SET_VOLUME) { //check for set volume help ret = dict_get_str (dict, "volname", &volname); if (volname && diff --git a/xlators/mgmt/glusterd/src/glusterd-quota.c b/xlators/mgmt/glusterd/src/glusterd-quota.c index cf23b6404..7f798ad26 100644 --- a/xlators/mgmt/glusterd/src/glusterd-quota.c +++ b/xlators/mgmt/glusterd/src/glusterd-quota.c @@ -474,7 +474,7 @@ glusterd_set_quota_limit (char *volname, char *path, char *hard_limit, new_limit.sl = hton64 (new_limit.sl); - ret = gf_string2bytesize (hard_limit, (uint64_t*)&new_limit.hl); + ret = gf_string2bytesize_uint64 (hard_limit, (uint64_t*)&new_limit.hl); if (ret) goto out; @@ -1400,13 +1400,13 @@ glusterd_op_stage_quota (dict_t *dict, char **op_errstr, dict_t *rsp_dict) "Faild to get hard-limit from dict"); goto out; } - ret = gf_string2bytesize (hard_limit_str, &hard_limit); + ret = gf_string2bytesize_uint64 (hard_limit_str, &hard_limit); if (ret) { gf_log (this->name, GF_LOG_ERROR, "Failed to convert hard-limit string to value"); goto out; } - if (hard_limit > INT64_MAX) { + if (hard_limit > UINT64_MAX) { ret = -1; ret = gf_asprintf (op_errstr, "Hard-limit %s is greater" " than %"PRId64"bytes. Please set a " diff --git a/xlators/mgmt/glusterd/src/glusterd-rpc-ops.c b/xlators/mgmt/glusterd/src/glusterd-rpc-ops.c index 27910d132..babd5a3be 100644 --- a/xlators/mgmt/glusterd/src/glusterd-rpc-ops.c +++ b/xlators/mgmt/glusterd/src/glusterd-rpc-ops.c @@ -1242,12 +1242,12 @@ int32_t glusterd_rpc_friend_add (call_frame_t *frame, xlator_t *this, void *data) { - gd1_mgmt_friend_req req = {{0},}; - int ret = 0; - glusterd_peerinfo_t *peerinfo = NULL; - glusterd_conf_t *priv = NULL; - glusterd_friend_sm_event_t *event = NULL; - dict_t *vols = NULL; + gd1_mgmt_friend_req req = {{0},}; + int ret = 0; + glusterd_peerinfo_t *peerinfo = NULL; + glusterd_conf_t *priv = NULL; + glusterd_friend_sm_event_t *event = NULL; + dict_t *peer_data = NULL; if (!frame || !this || !data) { @@ -1262,15 +1262,37 @@ glusterd_rpc_friend_add (call_frame_t *frame, xlator_t *this, peerinfo = event->peerinfo; - ret = glusterd_build_volume_dict (&vols); - if (ret) + ret = glusterd_add_volumes_to_export_dict (&peer_data); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, + "Unable to add list of volumes " + "in the peer_data dict for handshake"); goto out; + } + + if (priv->op_version >= GD_OP_VERSION_4) { + ret = glusterd_add_missed_snaps_to_export_dict (peer_data); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, + "Unable to add list of missed snapshots " + "in the peer_data dict for handshake"); + goto out; + } + + ret = glusterd_add_snapshots_to_export_dict (peer_data); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, + "Unable to add list of snapshots " + "in the peer_data dict for handshake"); + goto out; + } + } uuid_copy (req.uuid, MY_UUID); req.hostname = peerinfo->hostname; req.port = peerinfo->port; - ret = dict_allocate_and_serialize (vols, &req.vols.vols_val, + ret = dict_allocate_and_serialize (peer_data, &req.vols.vols_val, &req.vols.vols_len); if (ret) goto out; @@ -1284,8 +1306,8 @@ glusterd_rpc_friend_add (call_frame_t *frame, xlator_t *this, out: GF_FREE (req.vols.vols_val); - if (vols) - dict_unref (vols); + if (peer_data) + dict_unref (peer_data); gf_log ("glusterd", GF_LOG_DEBUG, "Returning %d", ret); return ret; diff --git a/xlators/mgmt/glusterd/src/glusterd-sm.c b/xlators/mgmt/glusterd/src/glusterd-sm.c index 3aaf359ac..7a8b2c94f 100644 --- a/xlators/mgmt/glusterd/src/glusterd-sm.c +++ b/xlators/mgmt/glusterd/src/glusterd-sm.c @@ -652,15 +652,14 @@ glusterd_ac_handle_friend_add_req (glusterd_friend_sm_event_t *event, void *ctx) glusterd_friend_update_ctx_t *new_ev_ctx = NULL; glusterd_friend_sm_event_t *new_event = NULL; glusterd_friend_sm_event_type_t event_type = GD_FRIEND_EVENT_NONE; + glusterd_conf_t *conf = NULL; int status = 0; int32_t op_ret = -1; int32_t op_errno = 0; - xlator_t *this = NULL; - glusterd_conf_t *priv = NULL; + xlator_t *this = NULL; this = THIS; - priv = this->private; - + GF_ASSERT (this); GF_ASSERT (ctx); ev_ctx = ctx; uuid_copy (uuid, ev_ctx->uuid); @@ -668,6 +667,9 @@ glusterd_ac_handle_friend_add_req (glusterd_friend_sm_event_t *event, void *ctx) GF_ASSERT (peerinfo); uuid_copy (peerinfo->uuid, ev_ctx->uuid); + conf = this->private; + GF_ASSERT (conf); + //Build comparison logic here. ret = glusterd_compare_friend_data (ev_ctx->vols, &status, peerinfo->hostname); @@ -683,6 +685,31 @@ glusterd_ac_handle_friend_add_req (glusterd_friend_sm_event_t *event, void *ctx) op_ret = -1; } + /* Compare missed_snapshot list with the peer * + * if volume comparison is successful */ + if ((op_ret == 0) && + (conf->op_version >= GD_OP_VERSION_4)) { + ret = glusterd_import_friend_missed_snap_list (ev_ctx->vols); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, + "Failed to import peer's " + "missed_snaps_list."); + event_type = GD_FRIEND_EVENT_LOCAL_RJT; + op_errno = GF_PROBE_MISSED_SNAP_CONFLICT; + op_ret = -1; + } + + ret = glusterd_compare_friend_snapshots (ev_ctx->vols, + peerinfo); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, + "Conflict in comparing peer's snapshots"); + event_type = GD_FRIEND_EVENT_LOCAL_RJT; + op_errno = GF_PROBE_SNAP_CONFLICT; + op_ret = -1; + } + } + ret = glusterd_friend_sm_new_event (event_type, &new_event); if (ret) { @@ -712,9 +739,9 @@ glusterd_ac_handle_friend_add_req (glusterd_friend_sm_event_t *event, void *ctx) // apply a deterministic function to decide via whom we should join the cluster if (strcmp(peerinfo->hostname, ev_ctx->hostname) > 0) { - stop_etcd(priv->etcd_pid); + stop_etcd(conf->etcd_pid); nuke_etcd_dir(); - priv->etcd_pid = start_etcd (uuid_utoa(MY_UUID), peerinfo->hostname); + conf->etcd_pid = start_etcd (uuid_utoa(MY_UUID), peerinfo->hostname); } out: diff --git a/xlators/mgmt/glusterd/src/glusterd-snapshot.c b/xlators/mgmt/glusterd/src/glusterd-snapshot.c index d11abee70..0e824a022 100644 --- a/xlators/mgmt/glusterd/src/glusterd-snapshot.c +++ b/xlators/mgmt/glusterd/src/glusterd-snapshot.c @@ -18,6 +18,14 @@ #include <sys/resource.h> #include <sys/statvfs.h> #include <sys/mount.h> +#include <signal.h> + + +#if !defined(__NetBSD__) && !defined(GF_DARWIN_HOST_OS) +#include <mntent.h> +#else +#include "mntent_compat.h" +#endif #include "globals.h" #include "compat.h" @@ -40,16 +48,70 @@ #include "cli1-xdr.h" #include "xdr-generic.h" -#ifdef GF_LINUX_HOST_OS -#include <mntent.h> -#endif +#include "lvm-defaults.h" char snap_mount_folder[PATH_MAX]; +/* Look for disconnected peers, for missed snap creates or deletes */ static int32_t glusterd_find_missed_snap (dict_t *rsp_dict, glusterd_volinfo_t *vol, - char *snap_uuid, struct list_head *peers, - int32_t op); + struct list_head *peers, int32_t op) +{ + int32_t brick_count = -1; + int32_t ret = -1; + xlator_t *this = NULL; + glusterd_peerinfo_t *peerinfo = NULL; + glusterd_brickinfo_t *brickinfo = NULL; + + this = THIS; + GF_ASSERT (this); + GF_ASSERT (rsp_dict); + GF_ASSERT (peers); + GF_ASSERT (vol); + + brick_count = 0; + list_for_each_entry (brickinfo, &vol->bricks, brick_list) { + if (!uuid_compare (brickinfo->uuid, MY_UUID)) { + /* If the brick belongs to the same node */ + brick_count++; + continue; + } + + list_for_each_entry (peerinfo, peers, uuid_list) { + if (uuid_compare (peerinfo->uuid, brickinfo->uuid)) { + /* If the brick doesnt belong to this peer */ + continue; + } + + /* Found peer who owns the brick, * + * if peer is not connected or not * + * friend add it to missed snap list */ + if (!(peerinfo->connected) || + (peerinfo->state.state != + GD_FRIEND_STATE_BEFRIENDED)) { + ret = glusterd_add_missed_snaps_to_dict + (rsp_dict, + vol, brickinfo, + brick_count + 1, + op); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, + "Failed to add missed snapshot " + "info for %s:%s in the " + "rsp_dict", brickinfo->hostname, + brickinfo->path); + goto out; + } + } + } + brick_count++; + } + + ret = 0; +out: + gf_log (this->name, GF_LOG_TRACE, "Returning %d", ret); + return ret; +} /* This function will restore a snapshot volumes * @@ -62,7 +124,6 @@ int glusterd_snapshot_restore (dict_t *dict, char **op_errstr, dict_t *rsp_dict) { int ret = -1; - char *volname = NULL; char *snapname = NULL; xlator_t *this = NULL; glusterd_volinfo_t *snap_volinfo = NULL; @@ -90,7 +151,7 @@ glusterd_snapshot_restore (dict_t *dict, char **op_errstr, dict_t *rsp_dict) snap = glusterd_find_snap_by_name (snapname); if (NULL == snap) { ret = gf_asprintf (op_errstr, "Snap (%s) not found", - snapname); + snapname); if (ret < 0) { goto out; } @@ -99,6 +160,7 @@ glusterd_snapshot_restore (dict_t *dict, char **op_errstr, dict_t *rsp_dict) goto out; } + /* TODO : As of now there is only volume in snapshot. * Change this when multiple volume snapshot is introduced */ @@ -116,7 +178,6 @@ glusterd_snapshot_restore (dict_t *dict, char **op_errstr, dict_t *rsp_dict) /* From origin glusterd check if * * any peers with snap bricks is down */ ret = glusterd_find_missed_snap (rsp_dict, snap_volinfo, - snap_volinfo->volname, &priv->peers, GF_SNAP_OPTION_TYPE_RESTORE); if (ret) { @@ -133,7 +194,7 @@ glusterd_snapshot_restore (dict_t *dict, char **op_errstr, dict_t *rsp_dict) * failure. */ gf_log (this->name, GF_LOG_ERROR, "Failed to restore " - "snap for %s volume", volname); + "snap for %s", snapname); goto out; } @@ -156,7 +217,7 @@ out: * @param rsp_dict response dictionary * @return Negative value on Failure and 0 in success */ -int +int32_t glusterd_snapshot_restore_prevalidate (dict_t *dict, char **op_errstr, dict_t *rsp_dict) { @@ -267,7 +328,7 @@ out: int snap_max_hard_limits_validate (dict_t *dict, char *volname, - uint64_t value, char **op_errstr) + uint64_t value, char **op_errstr) { char err_str[PATH_MAX] = ""; glusterd_conf_t *conf = NULL; @@ -449,7 +510,7 @@ glusterd_snap_create_pre_val_use_rsp_dict (dict_t *dst, dict_t *src) for (i = 0; i < volume_count; i++) { memset (snapbrckcnt, '\0', sizeof(snapbrckcnt)); ret = snprintf (snapbrckcnt, sizeof(snapbrckcnt) - 1, - "vol%ld_brickcount", i+1); + "vol%"PRId64"_brickcount", i+1); ret = dict_get_int64 (src, snapbrckcnt, &brick_count); if (ret) { gf_log (this->name, GF_LOG_TRACE, @@ -460,7 +521,7 @@ glusterd_snap_create_pre_val_use_rsp_dict (dict_t *dst, dict_t *src) for (j = 0; j < brick_count; j++) { /* Fetching data from source dict */ snprintf (key, sizeof(key) - 1, - "vol%ld.brickdir%ld", i+1, j); + "vol%"PRId64".brickdir%"PRId64, i+1, j); ret = dict_get_ptr (src, key, (void **)&snap_brick_dir); @@ -471,7 +532,7 @@ glusterd_snap_create_pre_val_use_rsp_dict (dict_t *dst, dict_t *src) } snprintf (key, sizeof(key) - 1, - "vol%ld.brick_snapdevice%ld", i+1, j); + "vol%"PRId64".brick_snapdevice%"PRId64, i+1, j); ret = dict_get_ptr (src, key, (void **)&snap_device); @@ -482,7 +543,7 @@ glusterd_snap_create_pre_val_use_rsp_dict (dict_t *dst, dict_t *src) } snprintf (snapbrckord, sizeof(snapbrckord) - 1, - "vol%ld.brick%ld.order", i+1, j); + "vol%"PRId64".brick%"PRId64".order", i+1, j); ret = dict_get_int64 (src, snapbrckord, &brick_order); if (ret) { @@ -493,7 +554,7 @@ glusterd_snap_create_pre_val_use_rsp_dict (dict_t *dst, dict_t *src) /* Adding the data in the dst dict */ snprintf (key, sizeof(key) - 1, - "vol%ld.brickdir%ld", i+1, brick_order); + "vol%"PRId64".brickdir%"PRId64, i+1, brick_order); tmpstr = gf_strdup (snap_brick_dir); if (!tmpstr) { @@ -511,7 +572,7 @@ glusterd_snap_create_pre_val_use_rsp_dict (dict_t *dst, dict_t *src) } snprintf (key, sizeof(key) - 1, - "vol%ld.brick_snapdevice%ld", + "vol%"PRId64".brick_snapdevice%"PRId64, i+1, brick_order); tmpstr = gf_strdup (snap_device); @@ -618,8 +679,8 @@ glusterd_snapshot_create_prevalidate (dict_t *dict, char **op_errstr, goto out; } if (volcount <= 0) { - snprintf (err_str, sizeof (err_str), "Invalid volume count %ld " - "supplied", volcount); + snprintf (err_str, sizeof (err_str), "Invalid volume count %"PRId64 + " supplied", volcount); ret = -1; goto out; } @@ -638,7 +699,7 @@ glusterd_snapshot_create_prevalidate (dict_t *dict, char **op_errstr, } for (i = 1; i <= volcount; i++) { - snprintf (key, sizeof (key), "volname%ld", i); + snprintf (key, sizeof (key), "volname%"PRId64, i); ret = dict_get_str (dict, key, &volname); if (ret) { snprintf (err_str, sizeof (err_str), @@ -690,7 +751,7 @@ glusterd_snapshot_create_prevalidate (dict_t *dict, char **op_errstr, goto out; } - snprintf (key, sizeof(key) - 1, "vol%ld_volid", i); + snprintf (key, sizeof(key) - 1, "vol%"PRId64"_volid", i); ret = dict_get_bin (dict, key, (void **)&snap_volid); if (ret) { gf_log (this->name, GF_LOG_ERROR, @@ -745,7 +806,7 @@ glusterd_snapshot_create_prevalidate (dict_t *dict, char **op_errstr, } snprintf (key, sizeof(key), - "vol%ld.brick_snapdevice%ld", i, + "vol%"PRId64".brick_snapdevice%"PRId64, i, brick_count); ret = dict_set_dynstr (rsp_dict, key, device); if (ret) { @@ -783,7 +844,7 @@ glusterd_snapshot_create_prevalidate (dict_t *dict, char **op_errstr, ret = -1; goto out; } - snprintf (key, sizeof(key), "vol%ld.brickdir%ld", i, + snprintf (key, sizeof(key), "vol%"PRId64".brickdir%"PRId64, i, brick_count); ret = dict_set_dynstr (rsp_dict, key, tmpstr); if (ret) { @@ -793,7 +854,7 @@ glusterd_snapshot_create_prevalidate (dict_t *dict, char **op_errstr, } tmpstr = NULL; - snprintf (key, sizeof(key) - 1, "vol%ld.brick%ld.order", + snprintf (key, sizeof(key) - 1, "vol%"PRId64".brick%"PRId64".order", i, brick_count); ret = dict_set_int64 (rsp_dict, key, brick_order); if (ret) { @@ -805,7 +866,7 @@ glusterd_snapshot_create_prevalidate (dict_t *dict, char **op_errstr, brick_count++; brick_order++; } - snprintf (key, sizeof(key) - 1, "vol%ld_brickcount", i); + snprintf (key, sizeof(key) - 1, "vol%"PRId64"_brickcount", i); ret = dict_set_int64 (rsp_dict, key, brick_count); if (ret) { gf_log (this->name, GF_LOG_ERROR, "Failed to set %s", @@ -1002,7 +1063,7 @@ glusterd_do_lvm_snapshot_remove (glusterd_volinfo_t *snap_vol, snprintf (msg, sizeof(msg), "remove snapshot of the brick %s:%s, " "device: %s", brickinfo->hostname, brickinfo->path, snap_device); - runner_add_args (&runner, "/sbin/lvremove", "-f", snap_device, NULL); + runner_add_args (&runner, LVM_REMOVE, "-f", snap_device, NULL); runner_log (&runner, "", GF_LOG_DEBUG, msg); ret = runner_run (&runner); @@ -1030,19 +1091,25 @@ glusterd_lvm_snapshot_remove (dict_t *rsp_dict, glusterd_volinfo_t *snap_vol) this = THIS; GF_ASSERT (this); - GF_ASSERT (rsp_dict); GF_ASSERT (snap_vol); - if (!snap_vol) { - gf_log (this->name, GF_LOG_ERROR, "snap volinfo is NULL"); + if ((snap_vol->is_snap_volume == _gf_false) && + (uuid_is_null (snap_vol->restored_from_snap))) { + gf_log (this->name, GF_LOG_DEBUG, + "Not a snap volume, or a restored snap volume."); + ret = 0; goto out; } brick_count = -1; list_for_each_entry (brickinfo, &snap_vol->bricks, brick_list) { brick_count++; - if (uuid_compare (brickinfo->uuid, MY_UUID)) + if (uuid_compare (brickinfo->uuid, MY_UUID)) { + gf_log (this->name, GF_LOG_DEBUG, + "%s:%s belongs to a different node", + brickinfo->hostname, brickinfo->path); continue; + } if (brickinfo->snap_status == -1) { gf_log (this->name, GF_LOG_INFO, @@ -1051,20 +1118,23 @@ glusterd_lvm_snapshot_remove (dict_t *rsp_dict, glusterd_volinfo_t *snap_vol) brickinfo->hostname, brickinfo->path, snap_vol->snapshot->snapname); - /* Adding missed delete to the dict */ - ret = glusterd_add_missed_snaps_to_dict + if (rsp_dict && + (snap_vol->is_snap_volume == _gf_true)) { + /* Adding missed delete to the dict */ + ret = glusterd_add_missed_snaps_to_dict (rsp_dict, - snap_vol->volname, + snap_vol, brickinfo, brick_count + 1, GF_SNAP_OPTION_TYPE_DELETE); - if (ret) { - gf_log (this->name, GF_LOG_ERROR, - "Failed to add missed snapshot info " - "for %s:%s in the rsp_dict", - brickinfo->hostname, - brickinfo->path); - goto out; + if (ret) { + gf_log (this->name, GF_LOG_ERROR, + "Failed to add missed snapshot " + "info for %s:%s in the " + "rsp_dict", brickinfo->hostname, + brickinfo->path); + goto out; + } } continue; @@ -1104,6 +1174,7 @@ glusterd_lvm_snapshot_remove (dict_t *rsp_dict, glusterd_volinfo_t *snap_vol) ret = 0; out: + gf_log (this->name, GF_LOG_TRACE, "Returning %d", ret); return ret; } @@ -2047,8 +2118,8 @@ glusterd_handle_snapshot_create (rpcsvc_request_t *req, glusterd_op_t op, goto out; } if (volcount <= 0) { - gf_log (this->name, GF_LOG_ERROR, "Invalid volume count %ld " - "supplied", volcount); + gf_log (this->name, GF_LOG_ERROR, "Invalid volume count %"PRId64 + " supplied", volcount); ret = -1; goto out; } @@ -2476,7 +2547,7 @@ glusterd_create_snap_object (dict_t *dict, dict_t *rsp_dict) } if (time_stamp <= 0) { ret = -1; - gf_log (this->name, GF_LOG_ERROR, "Invalid time-stamp: %ld", + gf_log (this->name, GF_LOG_ERROR, "Invalid time-stamp: %"PRId64, time_stamp); goto out; } @@ -2550,11 +2621,12 @@ out: /* Added missed_snap_entry to rsp_dict */ int32_t -glusterd_add_missed_snaps_to_dict (dict_t *rsp_dict, char *snap_uuid, +glusterd_add_missed_snaps_to_dict (dict_t *rsp_dict, + glusterd_volinfo_t *snap_vol, glusterd_brickinfo_t *brickinfo, int32_t brick_number, int32_t op) { - char *buf = NULL; + char *snap_uuid = NULL; char missed_snap_entry[PATH_MAX] = ""; char name_buf[PATH_MAX] = ""; int32_t missed_snap_count = -1; @@ -2564,20 +2636,20 @@ glusterd_add_missed_snaps_to_dict (dict_t *rsp_dict, char *snap_uuid, this = THIS; GF_ASSERT (this); GF_ASSERT (rsp_dict); - GF_ASSERT (snap_uuid); + GF_ASSERT (snap_vol); GF_ASSERT (brickinfo); - snprintf (missed_snap_entry, sizeof(missed_snap_entry), - "%s:%s=%d:%s:%d:%d", uuid_utoa(brickinfo->uuid), - snap_uuid, brick_number, brickinfo->path, op, - GD_MISSED_SNAP_PENDING); - - buf = gf_strdup (missed_snap_entry); - if (!buf) { + snap_uuid = gf_strdup (uuid_utoa (snap_vol->snapshot->snap_id)); + if (!snap_uuid) { ret = -1; goto out; } + snprintf (missed_snap_entry, sizeof(missed_snap_entry), + "%s:%s=%s:%d:%s:%d:%d", uuid_utoa(brickinfo->uuid), + snap_uuid, snap_vol->volname, brick_number, brickinfo->path, + op, GD_MISSED_SNAP_PENDING); + /* Fetch the missed_snap_count from the dict */ ret = dict_get_int32 (rsp_dict, "missed_snap_count", &missed_snap_count); @@ -2589,12 +2661,12 @@ glusterd_add_missed_snaps_to_dict (dict_t *rsp_dict, char *snap_uuid, /* Setting the missed_snap_entry in the rsp_dict */ snprintf (name_buf, sizeof(name_buf), "missed_snaps_%d", missed_snap_count); - ret = dict_set_dynstr (rsp_dict, name_buf, buf); + ret = dict_set_dynstr_with_alloc (rsp_dict, name_buf, + missed_snap_entry); if (ret) { gf_log (this->name, GF_LOG_ERROR, "Failed to set missed_snap_entry (%s) " - "in the rsp_dict.", buf); - GF_FREE (buf); + "in the rsp_dict.", missed_snap_entry); goto out; } missed_snap_count++; @@ -2610,6 +2682,9 @@ glusterd_add_missed_snaps_to_dict (dict_t *rsp_dict, char *snap_uuid, } out: + if (snap_uuid) + GF_FREE (snap_uuid); + gf_log (this->name, GF_LOG_TRACE, "Returning %d", ret); return ret; } @@ -2715,7 +2790,7 @@ glusterd_take_lvm_snapshot (glusterd_volinfo_t *snap_vol, /* Figuring out if setactivationskip flag is supported or not */ runinit (&runner); snprintf (msg, sizeof (msg), "running lvcreate help"); - runner_add_args (&runner, "/sbin/lvcreate", "--help", NULL); + runner_add_args (&runner, LVM_CREATE, "--help", NULL); runner_log (&runner, "", GF_LOG_DEBUG, msg); runner_redir (&runner, STDOUT_FILENO, RUN_PIPE); ret = runner_start (&runner); @@ -2744,11 +2819,11 @@ glusterd_take_lvm_snapshot (glusterd_volinfo_t *snap_vol, snprintf (msg, sizeof (msg), "taking snapshot of the brick %s:%s", brickinfo->hostname, brickinfo->path); if (match == _gf_true) - runner_add_args (&runner, "/sbin/lvcreate", "-s", device, + runner_add_args (&runner, LVM_CREATE, "-s", device, "--setactivationskip", "n", "--name", snap_vol->volname, NULL); else - runner_add_args (&runner, "/sbin/lvcreate", "-s", device, + runner_add_args (&runner, LVM_CREATE, "-s", device, "--name", snap_vol->volname, NULL); runner_log (&runner, "", GF_LOG_DEBUG, msg); ret = runner_start (&runner); @@ -2785,9 +2860,7 @@ glusterd_snap_brick_create (char *device, glusterd_volinfo_t *snap_volinfo, glusterd_conf_t *priv = NULL; char snap_brick_mount_path[PATH_MAX] = ""; char snap_brick_path[PATH_MAX] = ""; - char msg[1024] = ""; struct stat statbuf = {0, }; - runner_t runner = {0, }; this = THIS; priv = this->private; @@ -2818,25 +2891,12 @@ glusterd_snap_brick_create (char *device, glusterd_volinfo_t *snap_volinfo, MS_MGC_VAL, "nouuid"); But for now, mounting using runner apis. */ - runinit (&runner); - snprintf (msg, sizeof (msg), "mounting snapshot of the brick %s:%s", - original_brickinfo->hostname, original_brickinfo->path); - runner_add_args (&runner, "mount", "-o", "nouuid", device, - snap_brick_mount_path, NULL); - runner_log (&runner, "", GF_LOG_DEBUG, msg); - - /* let glusterd get blocked till snapshot is over */ - synclock_unlock (&priv->big_lock); - ret = runner_run (&runner); - synclock_lock (&priv->big_lock); + ret = glusterd_mount_lvm_snapshot (device, snap_brick_mount_path); if (ret) { - gf_log (this->name, GF_LOG_ERROR, "mounting the snapshot " - "logical device %s failed (error: %s)", device, - strerror (errno)); + gf_log (this->name, GF_LOG_ERROR, + "Failed to mount lvm snapshot."); goto out; - } else - gf_log (this->name, GF_LOG_DEBUG, "mounting the snapshot " - "logical device %s successful", device); + } ret = stat (snap_brick_path, &statbuf); if (ret) { @@ -2862,7 +2922,11 @@ out: if (ret) { gf_log (this->name, GF_LOG_WARNING, "unmounting the snap brick" " mount %s", snap_brick_mount_path); +#if !defined(GF_DARWIN_HOST_OS) umount (snap_brick_mount_path); +#else + unmount (snap_brick_mount_path, 0); +#endif } gf_log (this->name, GF_LOG_TRACE, "Returning %d", ret); @@ -2893,7 +2957,7 @@ glusterd_add_bricks_to_snap_volume (dict_t *dict, dict_t *rsp_dict, GF_ASSERT (snap_brickinfo); GF_ASSERT (snap_brick_dir); - snprintf (key, sizeof(key) - 1, "vol%ld.brickdir%d", volcount, + snprintf (key, sizeof(key) - 1, "vol%"PRId64".brickdir%d", volcount, brick_count); ret = dict_get_ptr (dict, key, (void **)snap_brick_dir); if (ret) { @@ -2938,7 +3002,7 @@ glusterd_add_bricks_to_snap_volume (dict_t *dict, dict_t *rsp_dict, if (add_missed_snap) { ret = glusterd_add_missed_snaps_to_dict (rsp_dict, - snap_vol->volname, + snap_vol, original_brickinfo, brick_count + 1, GF_SNAP_OPTION_TYPE_CREATE); @@ -2951,7 +3015,7 @@ glusterd_add_bricks_to_snap_volume (dict_t *dict, dict_t *rsp_dict, } } - snprintf (key, sizeof(key), "vol%ld.brick_snapdevice%d", + snprintf (key, sizeof(key), "vol%"PRId64".brick_snapdevice%d", volcount, brick_count); ret = dict_get_ptr (dict, key, (void **)&snap_device); if (ret) { @@ -3038,70 +3102,6 @@ out: return ret; } -/* Look for disconnected peers, for missed snap creates or deletes */ -static int32_t -glusterd_find_missed_snap (dict_t *rsp_dict, glusterd_volinfo_t *vol, - char *snap_uuid, struct list_head *peers, - int32_t op) -{ - int32_t brick_count = -1; - int32_t ret = -1; - xlator_t *this = NULL; - glusterd_peerinfo_t *peerinfo = NULL; - glusterd_brickinfo_t *brickinfo = NULL; - - this = THIS; - GF_ASSERT (this); - GF_ASSERT (rsp_dict); - GF_ASSERT (peers); - GF_ASSERT (vol); - GF_ASSERT (snap_uuid); - - brick_count = 0; - list_for_each_entry (brickinfo, &vol->bricks, brick_list) { - if (!uuid_compare (brickinfo->uuid, MY_UUID)) { - /* If the brick belongs to the same node */ - brick_count++; - continue; - } - - list_for_each_entry (peerinfo, peers, uuid_list) { - if (uuid_compare (peerinfo->uuid, brickinfo->uuid)) { - /* If the brick doesnt belong to this peer */ - continue; - } - - /* Found peer who owns the brick, * - * if peer is not connected or not * - * friend add it to missed snap list */ - if (!(peerinfo->connected) || - (peerinfo->state.state != - GD_FRIEND_STATE_BEFRIENDED)) { - ret = glusterd_add_missed_snaps_to_dict - (rsp_dict, - snap_uuid, - brickinfo, - brick_count + 1, - op); - if (ret) { - gf_log (this->name, GF_LOG_ERROR, - "Failed to add missed snapshot " - "info for %s:%s in the " - "rsp_dict", brickinfo->hostname, - brickinfo->path); - goto out; - } - } - } - brick_count++; - } - - ret = 0; -out: - gf_log (this->name, GF_LOG_TRACE, "Returning %d", ret); - return ret; -} - glusterd_volinfo_t * glusterd_do_snap_vol (glusterd_volinfo_t *origin_vol, glusterd_snap_t *snap, dict_t *dict, dict_t *rsp_dict, int64_t volcount) @@ -3129,7 +3129,7 @@ glusterd_do_snap_vol (glusterd_volinfo_t *origin_vol, glusterd_snap_t *snap, GF_ASSERT (rsp_dict); /* fetch username, password and vol_id from dict*/ - snprintf (key, sizeof(key), "volume%ld_username", volcount); + snprintf (key, sizeof(key), "volume%"PRId64"_username", volcount); ret = dict_get_str (dict, key, &username); if (ret) { gf_log (this->name, GF_LOG_ERROR, "Failed to get %s for " @@ -3137,7 +3137,7 @@ glusterd_do_snap_vol (glusterd_volinfo_t *origin_vol, glusterd_snap_t *snap, goto out; } - snprintf (key, sizeof(key), "volume%ld_password", volcount); + snprintf (key, sizeof(key), "volume%"PRId64"_password", volcount); ret = dict_get_str (dict, key, &password); if (ret) { gf_log (this->name, GF_LOG_ERROR, "Failed to get %s for " @@ -3145,7 +3145,7 @@ glusterd_do_snap_vol (glusterd_volinfo_t *origin_vol, glusterd_snap_t *snap, goto out; } - snprintf (key, sizeof(key) - 1, "vol%ld_volid", volcount); + snprintf (key, sizeof(key) - 1, "vol%"PRId64"_volid", volcount); ret = dict_get_bin (dict, key, (void **)&snap_volid); if (ret) { gf_log (this->name, GF_LOG_ERROR, @@ -3367,7 +3367,7 @@ glusterd_handle_snapshot_remove (rpcsvc_request_t *req, glusterd_op_t op, goto out; } - snprintf (key, sizeof (key), "volname%ld", volcount); + snprintf (key, sizeof (key), "volname%"PRId64, volcount); ret = dict_set_dynstr (dict, key, volname); if (ret) { gf_log (this->name, GF_LOG_ERROR, "Failed to set " @@ -3582,7 +3582,6 @@ glusterd_snapshot_remove_commit (dict_t *dict, char **op_errstr, /* From origin glusterd check if * * any peers with snap bricks is down */ ret = glusterd_find_missed_snap (rsp_dict, snap_volinfo, - snap_volinfo->volname, &priv->peers, GF_SNAP_OPTION_TYPE_DELETE); if (ret) { @@ -3697,7 +3696,14 @@ glusterd_snapshot_update_snaps_post_validate (dict_t *dict, char **op_errstr, goto out; } - ret = glusterd_store_update_missed_snaps (dict, missed_snap_count); + ret = glusterd_add_missed_snaps_to_list (dict, missed_snap_count); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, + "Failed to add missed snaps to list"); + goto out; + } + + ret = glusterd_store_update_missed_snaps (); if (ret) { gf_log (this->name, GF_LOG_ERROR, "Failed to update missed_snaps_list"); @@ -3771,7 +3777,7 @@ glusterd_snapshot_create_commit (dict_t *dict, char **op_errstr, } for (i = 1; i <= volcount; i++) { - snprintf (key, sizeof (key), "volname%ld", i); + snprintf (key, sizeof (key), "volname%"PRId64, i); ret = dict_get_str (dict, key, &volname); if (ret) { gf_log (this->name, GF_LOG_ERROR, @@ -3922,7 +3928,7 @@ snap_max_limits_display_commit (dict_t *rsp_dict, char *volname, soft_limit_value = (active_hard_limit * conf->snap_max_soft_limit) / 100; - snprintf (buf, sizeof(buf), "volume%ld-volname", count); + snprintf (buf, sizeof(buf), "volume%"PRIu64"-volname", count); ret = dict_set_str (rsp_dict, buf, volinfo->volname); if (ret) { snprintf (err_str, PATH_MAX, @@ -3931,7 +3937,7 @@ snap_max_limits_display_commit (dict_t *rsp_dict, char *volname, } snprintf (buf, sizeof(buf), - "volume%ld-snap-max-hard-limit", count); + "volume%"PRIu64"-snap-max-hard-limit", count); ret = dict_set_uint64 (rsp_dict, buf, snap_max_limit); if (ret) { snprintf (err_str, PATH_MAX, @@ -3940,7 +3946,7 @@ snap_max_limits_display_commit (dict_t *rsp_dict, char *volname, } snprintf (buf, sizeof(buf), - "volume%ld-active-hard-limit", count); + "volume%"PRIu64"-active-hard-limit", count); ret = dict_set_uint64 (rsp_dict, buf, active_hard_limit); if (ret) { @@ -3950,7 +3956,7 @@ snap_max_limits_display_commit (dict_t *rsp_dict, char *volname, } snprintf (buf, sizeof(buf), - "volume%ld-snap-max-soft-limit", count); + "volume%"PRIu64"-snap-max-soft-limit", count); ret = dict_set_uint64 (rsp_dict, buf, soft_limit_value); if (ret) { snprintf (err_str, PATH_MAX, @@ -3984,7 +3990,7 @@ snap_max_limits_display_commit (dict_t *rsp_dict, char *volname, soft_limit_value = (active_hard_limit * conf->snap_max_soft_limit) / 100; - snprintf (buf, sizeof(buf), "volume%ld-volname", count); + snprintf (buf, sizeof(buf), "volume%"PRIu64"-volname", count); ret = dict_set_str (rsp_dict, buf, volinfo->volname); if (ret) { snprintf (err_str, PATH_MAX, @@ -3993,7 +3999,7 @@ snap_max_limits_display_commit (dict_t *rsp_dict, char *volname, } snprintf (buf, sizeof(buf), - "volume%ld-snap-max-hard-limit", count); + "volume%"PRIu64"-snap-max-hard-limit", count); ret = dict_set_uint64 (rsp_dict, buf, snap_max_limit); if (ret) { snprintf (err_str, PATH_MAX, @@ -4002,7 +4008,7 @@ snap_max_limits_display_commit (dict_t *rsp_dict, char *volname, } snprintf (buf, sizeof(buf), - "volume%ld-active-hard-limit", count); + "volume%"PRIu64"-active-hard-limit", count); ret = dict_set_uint64 (rsp_dict, buf, active_hard_limit); if (ret) { snprintf (err_str, PATH_MAX, @@ -4011,7 +4017,7 @@ snap_max_limits_display_commit (dict_t *rsp_dict, char *volname, } snprintf (buf, sizeof(buf), - "volume%ld-snap-max-soft-limit", count); + "volume%"PRIu64"-snap-max-soft-limit", count); ret = dict_set_uint64 (rsp_dict, buf, soft_limit_value); if (ret) { snprintf (err_str, PATH_MAX, @@ -4196,7 +4202,7 @@ glusterd_get_brick_lvm_details (dict_t *rsp_dict, * for the above given command with separator ":", * The output will be "vgname:lvsize" */ - runner_add_args (&runner, "lvs", device, "--noheading", "-o", + runner_add_args (&runner, LVS, device, "--noheading", "-o", "vg_name,data_percent,lv_size", "--separator", ":", NULL); runner_redir (&runner, STDOUT_FILENO, RUN_PIPE); @@ -5320,6 +5326,30 @@ glusterd_free_snap_op (glusterd_snap_op_t *snap_op) } } +static inline void +glusterd_free_missed_snapinfo (glusterd_missed_snap_info *missed_snapinfo) +{ + glusterd_snap_op_t *snap_opinfo = NULL; + glusterd_snap_op_t *tmp = NULL; + + if (missed_snapinfo) { + list_for_each_entry_safe (snap_opinfo, tmp, + &missed_snapinfo->snap_ops, + snap_ops_list) { + glusterd_free_snap_op (snap_opinfo); + snap_opinfo = NULL; + } + + if (missed_snapinfo->node_uuid) + GF_FREE (missed_snapinfo->node_uuid); + + if (missed_snapinfo->snap_uuid) + GF_FREE (missed_snapinfo->snap_uuid); + + GF_FREE (missed_snapinfo); + } +} + /* Look for duplicates and accordingly update the list */ int32_t glusterd_update_missed_snap_entry (glusterd_missed_snap_info *missed_snapinfo, @@ -5337,6 +5367,13 @@ glusterd_update_missed_snap_entry (glusterd_missed_snap_info *missed_snapinfo, list_for_each_entry (snap_opinfo, &missed_snapinfo->snap_ops, snap_ops_list) { + /* If the entry is not for the same snap_vol_id + * then continue + */ + if (strcmp (snap_opinfo->snap_vol_id, + missed_snap_op->snap_vol_id)) + continue; + if ((!strcmp (snap_opinfo->brick_path, missed_snap_op->brick_path)) && (snap_opinfo->op == missed_snap_op->op)) { @@ -5348,8 +5385,10 @@ glusterd_update_missed_snap_entry (glusterd_missed_snap_info *missed_snapinfo, snap_opinfo->status = GD_MISSED_SNAP_DONE; gf_log (this->name, GF_LOG_INFO, "Updating missed snap status " - "for %s:%d:%s:%d as DONE", - missed_snapinfo->node_snap_info, + "for %s:%s=%s:%d:%s:%d as DONE", + missed_snapinfo->node_uuid, + missed_snapinfo->snap_uuid, + snap_opinfo->snap_vol_id, snap_opinfo->brick_num, snap_opinfo->brick_path, snap_opinfo->op); @@ -5362,15 +5401,19 @@ glusterd_update_missed_snap_entry (glusterd_missed_snap_info *missed_snapinfo, } else if ((snap_opinfo->brick_num == missed_snap_op->brick_num) && (snap_opinfo->op == GF_SNAP_OPTION_TYPE_CREATE) && - (missed_snap_op->op == - GF_SNAP_OPTION_TYPE_DELETE)) { + ((missed_snap_op->op == + GF_SNAP_OPTION_TYPE_DELETE) || + (missed_snap_op->op == + GF_SNAP_OPTION_TYPE_RESTORE))) { /* Optimizing create and delete entries for the same * brick and same node */ gf_log (this->name, GF_LOG_INFO, "Updating missed snap status " - "for %s:%d:%s:%d as DONE", - missed_snapinfo->node_snap_info, + "for %s:%s=%s:%d:%s:%d as DONE", + missed_snapinfo->node_uuid, + missed_snapinfo->snap_uuid, + snap_opinfo->snap_vol_id, snap_opinfo->brick_num, snap_opinfo->brick_path, snap_opinfo->op); @@ -5398,10 +5441,13 @@ out: /* Add new missed snap entry to the missed_snaps list. */ int32_t -glusterd_store_missed_snaps_list (char *missed_info, int32_t brick_num, - char *brick_path, int32_t snap_op, - int32_t snap_status) +glusterd_add_new_entry_to_list (char *missed_info, char *snap_vol_id, + int32_t brick_num, char *brick_path, + int32_t snap_op, int32_t snap_status) { + char *buf = NULL; + char *save_ptr = NULL; + char node_snap_info[PATH_MAX] = ""; int32_t ret = -1; glusterd_missed_snap_info *missed_snapinfo = NULL; glusterd_snap_op_t *missed_snap_op = NULL; @@ -5413,6 +5459,7 @@ glusterd_store_missed_snaps_list (char *missed_info, int32_t brick_num, this = THIS; GF_ASSERT(this); GF_ASSERT(missed_info); + GF_ASSERT(snap_vol_id); GF_ASSERT(brick_path); priv = this->private; @@ -5428,6 +5475,11 @@ glusterd_store_missed_snaps_list (char *missed_info, int32_t brick_num, goto out; } + missed_snap_op->snap_vol_id = gf_strdup(snap_vol_id); + if (!missed_snap_op->snap_vol_id) { + ret = -1; + goto out; + } missed_snap_op->brick_path = gf_strdup(brick_path); if (!missed_snap_op->brick_path) { ret = -1; @@ -5440,8 +5492,10 @@ glusterd_store_missed_snaps_list (char *missed_info, int32_t brick_num, /* Look for other entries for the same node and same snap */ list_for_each_entry (missed_snapinfo, &priv->missed_snaps_list, missed_snaps) { - if (!strcmp (missed_snapinfo->node_snap_info, - missed_info)) { + snprintf (node_snap_info, sizeof(node_snap_info), + "%s:%s", missed_snapinfo->node_uuid, + missed_snapinfo->snap_uuid); + if (!strcmp (node_snap_info, missed_info)) { /* Found missed snapshot info for * * the same node and same snap */ match = _gf_true; @@ -5458,8 +5512,24 @@ glusterd_store_missed_snaps_list (char *missed_info, int32_t brick_num, goto out; } free_missed_snap_info = _gf_true; - missed_snapinfo->node_snap_info = gf_strdup(missed_info); - if (!missed_snapinfo->node_snap_info) { + buf = strtok_r (missed_info, ":", &save_ptr); + if (!buf) { + ret = -1; + goto out; + } + missed_snapinfo->node_uuid = gf_strdup(buf); + if (!missed_snapinfo->node_uuid) { + ret = -1; + goto out; + } + + buf = strtok_r (NULL, ":", &save_ptr); + if (!buf) { + ret = -1; + goto out; + } + missed_snapinfo->snap_uuid = gf_strdup(buf); + if (!missed_snapinfo->snap_uuid) { ret = -1; goto out; } @@ -5486,12 +5556,8 @@ out: glusterd_free_snap_op (missed_snap_op); if (missed_snapinfo && - (free_missed_snap_info == _gf_true)) { - if (missed_snapinfo->node_snap_info) - GF_FREE (missed_snapinfo->node_snap_info); - - GF_FREE (missed_snapinfo); - } + (free_missed_snap_info == _gf_true)) + glusterd_free_missed_snapinfo (missed_snapinfo); } gf_log (this->name, GF_LOG_TRACE, "Returning %d", ret); @@ -5507,6 +5573,7 @@ glusterd_add_missed_snaps_to_list (dict_t *dict, int32_t missed_snap_count) char *save_ptr = NULL; char *nodeid = NULL; char *snap_uuid = NULL; + char *snap_vol_id = NULL; char *brick_path = NULL; char missed_info[PATH_MAX] = ""; char name_buf[PATH_MAX] = ""; @@ -5553,13 +5620,14 @@ glusterd_add_missed_snaps_to_list (dict_t *dict, int32_t missed_snap_count) */ nodeid = strtok_r (tmp, ":", &save_ptr); snap_uuid = strtok_r (NULL, "=", &save_ptr); + snap_vol_id = strtok_r (NULL, ":", &save_ptr); brick_num = atoi(strtok_r (NULL, ":", &save_ptr)); brick_path = strtok_r (NULL, ":", &save_ptr); snap_op = atoi(strtok_r (NULL, ":", &save_ptr)); snap_status = atoi(strtok_r (NULL, ":", &save_ptr)); if (!nodeid || !snap_uuid || !brick_path || - brick_num < 1 || snap_op < 1 || + !snap_vol_id || brick_num < 1 || snap_op < 1 || snap_status < 1) { gf_log (this->name, GF_LOG_ERROR, "Invalid missed_snap_entry"); @@ -5570,11 +5638,12 @@ glusterd_add_missed_snaps_to_list (dict_t *dict, int32_t missed_snap_count) snprintf (missed_info, sizeof(missed_info), "%s:%s", nodeid, snap_uuid); - ret = glusterd_store_missed_snaps_list (missed_info, - brick_num, - brick_path, - snap_op, - snap_status); + ret = glusterd_add_new_entry_to_list (missed_info, + snap_vol_id, + brick_num, + brick_path, + snap_op, + snap_status); if (ret) { gf_log (this->name, GF_LOG_ERROR, "Failed to store missed snaps_list"); @@ -5585,6 +5654,7 @@ glusterd_add_missed_snaps_to_list (dict_t *dict, int32_t missed_snap_count) tmp = NULL; } + ret = 0; out: if (tmp) GF_FREE (tmp); @@ -5592,3 +5662,126 @@ out: gf_log (this->name, GF_LOG_TRACE, "Returning %d", ret); return ret; } + +/* This function will restore origin volume to it's snap. + * The restore operation will simply replace the Gluster origin + * volume with the snap volume. + * TODO: Multi-volume delete to be done. + * Cleanup in case of restore failure is pending. + * + * @param orig_vol volinfo of origin volume + * @param snap_vol volinfo of snapshot volume + * + * @return 0 on success and negative value on error + */ +int +gd_restore_snap_volume (dict_t *rsp_dict, + glusterd_volinfo_t *orig_vol, + glusterd_volinfo_t *snap_vol) +{ + int ret = -1; + glusterd_volinfo_t *new_volinfo = NULL; + glusterd_snap_t *snap = NULL; + xlator_t *this = NULL; + glusterd_conf_t *conf = NULL; + glusterd_volinfo_t *temp_volinfo = NULL; + glusterd_volinfo_t *voliter = NULL; + + this = THIS; + GF_ASSERT (this); + GF_ASSERT (rsp_dict); + conf = this->private; + GF_ASSERT (conf); + + GF_VALIDATE_OR_GOTO (this->name, orig_vol, out); + GF_VALIDATE_OR_GOTO (this->name, snap_vol, out); + snap = snap_vol->snapshot; + GF_VALIDATE_OR_GOTO (this->name, snap, out); + + /* Snap volume must be stoped before performing the + * restore operation. + */ + ret = glusterd_stop_volume (snap_vol); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, "Failed to stop " + "snap volume"); + goto out; + } + + /* Create a new volinfo for the restored volume */ + ret = glusterd_volinfo_dup (snap_vol, &new_volinfo, _gf_true); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, "Failed to create volinfo"); + goto out; + } + + /* Following entries need to be derived from origin volume. */ + strcpy (new_volinfo->volname, orig_vol->volname); + uuid_copy (new_volinfo->volume_id, orig_vol->volume_id); + new_volinfo->snap_count = orig_vol->snap_count; + new_volinfo->snap_max_hard_limit = orig_vol->snap_max_hard_limit; + uuid_copy (new_volinfo->restored_from_snap, + snap_vol->snapshot->snap_id); + + /* Bump the version of the restored volume, so that nodes * + * which are done can sync during handshake */ + new_volinfo->version = orig_vol->version; + + list_for_each_entry_safe (voliter, temp_volinfo, + &orig_vol->snap_volumes, snapvol_list) { + list_add_tail (&voliter->snapvol_list, + &new_volinfo->snap_volumes); + } + /* Copy the snap vol info to the new_volinfo.*/ + ret = glusterd_snap_volinfo_restore (rsp_dict, new_volinfo, snap_vol); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, "Failed to restore snap"); + (void)glusterd_volinfo_delete (new_volinfo); + goto out; + } + + ret = glusterd_lvm_snapshot_remove (rsp_dict, orig_vol); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, "Failed to remove " + "LVM backend"); + (void)glusterd_volinfo_delete (new_volinfo); + goto out; + } + + /* New volinfo always shows the status as created. Therefore + * set the status to the original volume's status. */ + glusterd_set_volume_status (new_volinfo, orig_vol->status); + + /* Once the new_volinfo is completely constructed then delete + * the orinal volinfo + */ + ret = glusterd_volinfo_delete (orig_vol); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, "Failed to delete volinfo"); + (void)glusterd_volinfo_delete (new_volinfo); + goto out; + } + + list_add_tail (&new_volinfo->vol_list, &conf->volumes); + + /* Now delete the snap entry. As a first step delete the snap + * volume information stored in store. */ + ret = glusterd_snap_remove (rsp_dict, snap, _gf_false, _gf_true); + if (ret) { + gf_log (this->name, GF_LOG_WARNING, "Failed to delete " + "snap %s", snap->snapname); + goto out; + } + + ret = glusterd_store_volinfo (new_volinfo, + GLUSTERD_VOLINFO_VER_AC_INCREMENT); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, "Failed to store volinfo"); + goto out; + } + + ret = 0; +out: + + return ret; +} diff --git a/xlators/mgmt/glusterd/src/glusterd-store.c b/xlators/mgmt/glusterd/src/glusterd-store.c index e28a30c5a..afbc8ff35 100644 --- a/xlators/mgmt/glusterd/src/glusterd-store.c +++ b/xlators/mgmt/glusterd/src/glusterd-store.c @@ -44,6 +44,7 @@ #include <sys/resource.h> #include <inttypes.h> #include <dirent.h> +#include <mntent.h> void glusterd_replace_slash_with_hyphen (char *str) @@ -525,10 +526,13 @@ int _storeopts (dict_t *this, char *key, data_t *value, void *data) int32_t glusterd_volume_exclude_options_write (int fd, glusterd_volinfo_t *volinfo) { - char *str = NULL; - char buf[PATH_MAX] = {0,}; - int32_t ret = -1; + char *str = NULL; + char buf[PATH_MAX] = ""; + int32_t ret = -1; + xlator_t *this = NULL; + this = THIS; + GF_ASSERT (this); GF_ASSERT (fd > 0); GF_ASSERT (volinfo); @@ -576,7 +580,7 @@ glusterd_volume_exclude_options_write (int fd, glusterd_volinfo_t *volinfo) snprintf (buf, sizeof (buf), "%s", volinfo->parent_volname); ret = gf_store_save_value (fd, GLUSTERD_STORE_KEY_PARENT_VOLNAME, buf); if (ret) { - gf_log (THIS->name, GF_LOG_ERROR, "Failed to store " + gf_log (this->name, GF_LOG_ERROR, "Failed to store " GLUSTERD_STORE_KEY_PARENT_VOLNAME); goto out; } @@ -620,11 +624,11 @@ glusterd_volume_exclude_options_write (int fd, glusterd_volinfo_t *volinfo) goto out; } - snprintf (buf, sizeof (buf), "%d", volinfo->is_volume_restored); - ret = gf_store_save_value (fd, GLUSTERD_STORE_KEY_VOL_IS_RESTORED, buf); + ret = gf_store_save_value (fd, GLUSTERD_STORE_KEY_VOL_RESTORED_SNAP, + uuid_utoa (volinfo->restored_from_snap)); if (ret) { - gf_log (THIS->name, GF_LOG_ERROR, - "Unable to write is_volume_restored"); + gf_log (this->name, GF_LOG_ERROR, + "Unable to write restored_from_snap"); goto out; } @@ -632,14 +636,14 @@ glusterd_volume_exclude_options_write (int fd, glusterd_volinfo_t *volinfo) ret = gf_store_save_value (fd, GLUSTERD_STORE_KEY_SNAP_MAX_HARD_LIMIT, buf); if (ret) { - gf_log (THIS->name, GF_LOG_ERROR, + gf_log (this->name, GF_LOG_ERROR, "Unable to write snap-max-hard-limit"); goto out; } out: if (ret) - gf_log (THIS->name, GF_LOG_ERROR, "Unable to write volume " + gf_log (this->name, GF_LOG_ERROR, "Unable to write volume " "values for %s", volinfo->volname); return ret; } @@ -673,7 +677,7 @@ glusterd_store_create_volume_dir (glusterd_volinfo_t *volinfo) return ret; } -static int32_t +int32_t glusterd_store_create_snap_dir (glusterd_snap_t *snap) { int32_t ret = -1; @@ -2417,9 +2421,12 @@ glusterd_store_update_volinfo (glusterd_volinfo_t *volinfo) } else if (!strncmp (key, GLUSTERD_STORE_KEY_SNAP_MAX_HARD_LIMIT, strlen (GLUSTERD_STORE_KEY_SNAP_MAX_HARD_LIMIT))) { volinfo->snap_max_hard_limit = (uint64_t) atoll (value); - } else if (!strncmp (key, GLUSTERD_STORE_KEY_VOL_IS_RESTORED, - strlen (GLUSTERD_STORE_KEY_VOL_IS_RESTORED))) { - volinfo->is_volume_restored = atoi (value); + } else if (!strncmp (key, GLUSTERD_STORE_KEY_VOL_RESTORED_SNAP, + strlen (GLUSTERD_STORE_KEY_VOL_RESTORED_SNAP))) { + ret = uuid_parse (value, volinfo->restored_from_snap); + if (ret) + gf_log (this->name, GF_LOG_WARNING, + "failed to parse restored snap's uuid"); } else if (!strncmp (key, GLUSTERD_STORE_KEY_PARENT_VOLNAME, strlen (GLUSTERD_STORE_KEY_PARENT_VOLNAME))) { strncpy (volinfo->parent_volname, value, sizeof(volinfo->parent_volname) - 1); @@ -2615,7 +2622,7 @@ out: return volinfo; } -inline void +static inline void glusterd_store_set_options_path (glusterd_conf_t *conf, char *path, size_t len) { snprintf (path, len, "%s/options", conf->workdir); @@ -2786,6 +2793,209 @@ out: return ret; } +/* Figure out the brick mount path, from the brick path */ +int32_t +glusterd_find_brick_mount_path (char *brick_path, int32_t brick_count, + char **brick_mount_path) +{ + char brick_num[PATH_MAX] = ""; + char *ptr = NULL; + int32_t ret = -1; + xlator_t *this = NULL; + + this = THIS; + GF_ASSERT (this); + GF_ASSERT (brick_path); + GF_ASSERT (brick_mount_path); + + *brick_mount_path = gf_strdup (brick_path); + if (!*brick_mount_path) { + ret = -1; + goto out; + } + + snprintf (brick_num, sizeof(brick_num), "brick%d", brick_count); + + /* Finding the pointer to the end of + * /var/run/gluster/snaps/<snap-uuid> + */ + ptr = strstr (*brick_mount_path, brick_num); + if (!ptr) { + /* Snapshot bricks must have brick num as part + * of the brickpath + */ + gf_log (this->name, GF_LOG_ERROR, + "Invalid brick path(%s)", brick_path); + ret = -1; + goto out; + } + + /* Moving the pointer to the end of + * /var/run/gluster/snaps/<snap-uuid>/<brick_num> + * and assigning '\0' to it. + */ + ptr += strlen(brick_num); + *ptr = '\0'; + + ret = 0; +out: + if (ret && *brick_mount_path) { + GF_FREE (*brick_mount_path); + *brick_mount_path = NULL; + } + gf_log (this->name, GF_LOG_TRACE, "Returning with %d", ret); + return ret; +} + +/* Check if brick_mount_path is already mounted. If not, mount the device_path + * at the brick_mount_path + */ +int32_t +glusterd_mount_brick_paths (char *brick_mount_path, char *device_path) +{ + FILE *mtab = NULL; + int32_t ret = -1; + runner_t runner = {0, }; + struct mntent *entry = NULL; + xlator_t *this = NULL; + glusterd_conf_t *priv = NULL; + + this = THIS; + GF_ASSERT (this); + GF_ASSERT (brick_mount_path); + GF_ASSERT (device_path); + + priv = this->private; + GF_ASSERT (priv); + + /* Check if the brick_mount_path is already mounted */ + entry = glusterd_get_mnt_entry_info (brick_mount_path, mtab); + if (entry) { + gf_log (this->name, GF_LOG_INFO, + "brick_mount_path (%s) already mounted.", + brick_mount_path); + ret = 0; + goto out; + } + + /* TODO RHEL 6.5 has the logical volumes inactive by default + * on reboot. Hence activating the logical vol. Check behaviour + * on other systems + */ + /* Activate the snapshot */ + runinit (&runner); + runner_add_args (&runner, "lvchange", "-ay", device_path, + NULL); + ret = runner_run (&runner); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, + "Failed to activate %s. Error: %s", + device_path, strerror(errno)); + goto out; + } else + gf_log (this->name, GF_LOG_DEBUG, + "Activating %s successful", device_path); + + /* Mount the snapshot */ + ret = glusterd_mount_lvm_snapshot (device_path, brick_mount_path); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, + "Failed to mount lvm snapshot."); + goto out; + } + +out: + if (mtab) + endmntent (mtab); + gf_log (this->name, GF_LOG_TRACE, "Returning with %d", ret); + return ret; +} + +static int32_t +glusterd_store_recreate_brick_mounts (glusterd_volinfo_t *volinfo) +{ + char *brick_mount_path = NULL; + glusterd_brickinfo_t *brickinfo = NULL; + int32_t ret = -1; + int32_t brick_count = -1; + struct stat st_buf = {0, }; + xlator_t *this = NULL; + + this = THIS; + GF_ASSERT (this); + GF_ASSERT (volinfo); + + brick_count = 0; + list_for_each_entry (brickinfo, &volinfo->bricks, brick_list) { + brick_count++; + /* If the brick is not of this node, or its + * snapshot is pending, or the brick is not + * a snapshotted brick, we continue + */ + if ((uuid_compare (brickinfo->uuid, MY_UUID)) || + (brickinfo->snap_status == -1) || + (strlen(brickinfo->device_path) == 0)) + continue; + + /* Fetch the brick mount path from the brickinfo->path */ + ret = glusterd_find_brick_mount_path (brickinfo->path, + brick_count, + &brick_mount_path); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, + "Failed to find brick_mount_path for %s", + brickinfo->path); + goto out; + } + + /* Check if the brickinfo path is present. + * If not create the brick_mount_path */ + ret = lstat (brickinfo->path, &st_buf); + if (ret) { + if (errno == ENOENT) { + ret = mkdir_p (brick_mount_path, 0777, + _gf_true); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, + "Failed to create %s. " + "Error: %s", brick_mount_path, + strerror (errno)); + goto out; + } + } else { + gf_log (this->name, GF_LOG_ERROR, + "Brick Path(%s) not valid. " + "Error: %s", brickinfo->path, + strerror(errno)); + goto out; + } + } + + /* Check if brick_mount_path is already mounted. + * If not, mount the device_path at the brick_mount_path */ + ret = glusterd_mount_brick_paths (brick_mount_path, + brickinfo->device_path); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, + "Failed to mount brick_mount_path"); + goto out; + } + + if (brick_mount_path) { + GF_FREE (brick_mount_path); + brick_mount_path = NULL; + } + } + + ret = 0; +out: + if (ret && brick_mount_path) + GF_FREE (brick_mount_path); + + gf_log (this->name, GF_LOG_TRACE, "Returning with %d", ret); + return ret; +} + int32_t glusterd_resolve_snap_bricks (xlator_t *this, glusterd_snap_t *snap) { @@ -2905,25 +3115,16 @@ out: int32_t glusterd_store_retrieve_snap (char *snapname) { - int32_t ret = -1; - dict_t *dict = NULL; - glusterd_snap_t *snap = NULL; - glusterd_conf_t *priv = NULL; - xlator_t *this = NULL; + int32_t ret = -1; + glusterd_snap_t *snap = NULL; + glusterd_conf_t *priv = NULL; + xlator_t *this = NULL; this = THIS; priv = this->private; GF_ASSERT (priv); GF_ASSERT (snapname); - dict = dict_new(); - if (!dict) { - gf_log (this->name, GF_LOG_ERROR, - "Failed to create dict"); - ret = -1; - goto out; - } - snap = glusterd_new_snap_object (); if (!snap) { gf_log (this->name, GF_LOG_ERROR, "Failed to create " @@ -2946,34 +3147,6 @@ glusterd_store_retrieve_snap (char *snapname) goto out; } - /* Unlike bricks of normal volumes which are resolved at the end of - the glusterd restore, the bricks belonging to the snap volumes of - each snap should be resolved as part of snapshot restore itself. - Because if the snapshot has to be removed, then resolving bricks - helps glusterd in understanding what all bricks have its own uuid - and killing those bricks. - */ - ret = glusterd_resolve_snap_bricks (this, snap); - if (ret) - gf_log (this->name, GF_LOG_WARNING, "resolving the snap bricks" - " failed (snap: %s)", snap?snap->snapname:""); - - /* When the snapshot command from cli is received, the on disk and - in memory structures for the snapshot are created (with the status) - being marked as GD_SNAP_STATUS_INIT. Once the backend snapshot is - taken, the status is changed to GD_SNAP_STATUS_IN_USE. If glusterd - dies after taking the backend snapshot, but before updating the - status, then when glusterd comes up, it should treat that snapshot - as a failed snapshot and clean it up. - */ - if (snap->snap_status != GD_SNAP_STATUS_IN_USE) { - ret = glusterd_snap_remove (dict, snap, _gf_true, _gf_true); - if (ret) - gf_log (this->name, GF_LOG_WARNING, "failed to remove" - " the snapshot %s", snap->snapname); - goto out; - } - /* TODO: list_add_order can do 'N-square' comparisions and is not efficient. Find a better solution to store the snap in order */ @@ -2981,9 +3154,6 @@ glusterd_store_retrieve_snap (char *snapname) glusterd_compare_snap_time); out: - if (dict) - dict_unref (dict); - gf_log (this->name, GF_LOG_TRACE, "Returning with %d", ret); return ret; } @@ -2994,6 +3164,7 @@ glusterd_store_retrieve_missed_snaps_list (xlator_t *this) { char buf[PATH_MAX] = ""; char path[PATH_MAX] = ""; + char *snap_vol_id = NULL; char *missed_node_info = NULL; char *brick_path = NULL; char *value = NULL; @@ -3048,12 +3219,13 @@ glusterd_store_retrieve_missed_snaps_list (xlator_t *this) } /* Fetch the brick_num, brick_path, snap_op and snap status */ - brick_num = atoi(strtok_r (value, ":", &save_ptr)); + snap_vol_id = strtok_r (value, ":", &save_ptr); + brick_num = atoi(strtok_r (NULL, ":", &save_ptr)); brick_path = strtok_r (NULL, ":", &save_ptr); snap_op = atoi(strtok_r (NULL, ":", &save_ptr)); snap_status = atoi(strtok_r (NULL, ":", &save_ptr)); - if (!missed_node_info || !brick_path || + if (!missed_node_info || !brick_path || !snap_vol_id || brick_num < 1 || snap_op < 1 || snap_status < 1) { gf_log (this->name, GF_LOG_ERROR, @@ -3062,11 +3234,12 @@ glusterd_store_retrieve_missed_snaps_list (xlator_t *this) goto out; } - ret = glusterd_store_missed_snaps_list (missed_node_info, - brick_num, - brick_path, - snap_op, - snap_status); + ret = glusterd_add_new_entry_to_list (missed_node_info, + snap_vol_id, + brick_num, + brick_path, + snap_op, + snap_status); if (ret) { gf_log (this->name, GF_LOG_ERROR, "Failed to store missed snaps_list"); @@ -3145,6 +3318,7 @@ out: int32_t glusterd_store_write_missed_snapinfo (int32_t fd) { + char key[PATH_MAX] = ""; char value[PATH_MAX] = ""; int32_t ret = -1; glusterd_conf_t *priv = NULL; @@ -3164,14 +3338,15 @@ glusterd_store_write_missed_snapinfo (int32_t fd) list_for_each_entry (snap_opinfo, &missed_snapinfo->snap_ops, snap_ops_list) { - snprintf (value, sizeof(value), "%d:%s:%d:%d", + snprintf (key, sizeof(key), "%s:%s", + missed_snapinfo->node_uuid, + missed_snapinfo->snap_uuid); + snprintf (value, sizeof(value), "%s:%d:%s:%d:%d", + snap_opinfo->snap_vol_id, snap_opinfo->brick_num, snap_opinfo->brick_path, snap_opinfo->op, snap_opinfo->status); - ret = gf_store_save_value - (fd, - missed_snapinfo->node_snap_info, - value); + ret = gf_store_save_value (fd, key, value); if (ret) { gf_log (this->name, GF_LOG_ERROR, "Failed to write missed snapinfo"); @@ -3189,7 +3364,7 @@ out: /* Adds the missed snap entries to the in-memory conf->missed_snap_list * * and writes them to disk */ int32_t -glusterd_store_update_missed_snaps (dict_t *dict, int32_t missed_snap_count) +glusterd_store_update_missed_snaps () { int32_t fd = -1; int32_t ret = -1; @@ -3198,17 +3373,10 @@ glusterd_store_update_missed_snaps (dict_t *dict, int32_t missed_snap_count) this = THIS; GF_ASSERT(this); - GF_ASSERT(dict); priv = this->private; GF_ASSERT (priv); - if (missed_snap_count < 1) { - gf_log (this->name, GF_LOG_DEBUG, "No missed snaps"); - ret = 0; - goto out; - } - ret = glusterd_store_create_missed_snaps_list_shandle_on_absence (); if (ret) { gf_log (this->name, GF_LOG_ERROR, "Unable to obtain " @@ -3224,13 +3392,6 @@ glusterd_store_update_missed_snaps (dict_t *dict, int32_t missed_snap_count) goto out; } - ret = glusterd_add_missed_snaps_to_list (dict, missed_snap_count); - if (ret) { - gf_log (this->name, GF_LOG_ERROR, - "Failed to add missed snaps to list"); - goto out; - } - ret = glusterd_store_write_missed_snapinfo (fd); if (ret) { gf_log (this->name, GF_LOG_ERROR, @@ -3619,19 +3780,147 @@ out: return ret; } +static int32_t +glusterd_recreate_vol_brick_mounts (xlator_t *this, + glusterd_volinfo_t *volinfo) +{ + int32_t ret = 0; + glusterd_conf_t *priv = NULL; + glusterd_brickinfo_t *brickinfo = NULL; + + GF_ASSERT (this); + priv = this->private; + GF_ASSERT (priv); + + list_for_each_entry (brickinfo, &volinfo->bricks, brick_list) { + ret = glusterd_store_recreate_brick_mounts (volinfo); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, + "Failed to recreate brick mounts " + "for %s", volinfo->volname); + goto out; + } + } + +out: + gf_log (this->name, GF_LOG_TRACE, "Returning with %d", ret); + return ret; +} + +/* Bricks for snap volumes are hosted at /var/run/gluster/snaps + * When a volume is restored, it points to the bricks of the snap + * volume it was restored from. Hence on a node restart these + * paths need to be recreated and re-mounted + */ +int32_t +glusterd_recreate_all_snap_brick_mounts (xlator_t *this) +{ + int32_t ret = 0; + glusterd_conf_t *priv = NULL; + glusterd_volinfo_t *volinfo = NULL; + glusterd_snap_t *snap = NULL; + + GF_ASSERT (this); + priv = this->private; + GF_ASSERT (priv); + + /* Recreate bricks of volumes restored from snaps */ + list_for_each_entry (volinfo, &priv->volumes, vol_list) { + /* If the volume is not a restored volume then continue */ + if (uuid_is_null (volinfo->restored_from_snap)) + continue; + + ret = glusterd_recreate_vol_brick_mounts (this, volinfo); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, + "Failed to recreate brick mounts " + "for %s", volinfo->volname); + goto out; + } + } + + /* Recreate bricks of snapshot volumes */ + list_for_each_entry (snap, &priv->snapshots, snap_list) { + list_for_each_entry (volinfo, &snap->volumes, vol_list) { + ret = glusterd_recreate_vol_brick_mounts (this, + volinfo); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, + "Failed to recreate brick mounts " + "for %s", snap->snapname); + goto out; + } + } + } + +out: + gf_log (this->name, GF_LOG_TRACE, "Returning with %d", ret); + return ret; +} + +/* When the snapshot command from cli is received, the on disk and + * in memory structures for the snapshot are created (with the status) + * being marked as GD_SNAP_STATUS_INIT. Once the backend snapshot is + * taken, the status is changed to GD_SNAP_STATUS_IN_USE. If glusterd + * dies after taking the backend snapshot, but before updating the + * status, then when glusterd comes up, it should treat that snapshot + * as a failed snapshot and clean it up. + */ +int32_t +glusterd_snap_cleanup (xlator_t *this) +{ + dict_t *dict = NULL; + int32_t ret = 0; + glusterd_conf_t *priv = NULL; + glusterd_snap_t *snap = NULL; + + GF_ASSERT (this); + priv = this->private; + GF_ASSERT (priv); + + dict = dict_new(); + if (!dict) { + gf_log (this->name, GF_LOG_ERROR, + "Failed to create dict"); + ret = -1; + goto out; + } + + list_for_each_entry (snap, &priv->snapshots, snap_list) { + if (snap->snap_status != GD_SNAP_STATUS_IN_USE) { + ret = glusterd_snap_remove (dict, snap, + _gf_true, _gf_true); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, + "Failed to remove the snapshot %s", + snap->snapname); + goto out; + } + } + } +out: + if (dict) + dict_unref (dict); + + gf_log (this->name, GF_LOG_TRACE, "Returning with %d", ret); + return ret; +} + int32_t glusterd_resolve_all_bricks (xlator_t *this) { - int32_t ret = 0; - glusterd_conf_t *priv = NULL; - glusterd_volinfo_t *volinfo = NULL; + int32_t ret = 0; + glusterd_conf_t *priv = NULL; + glusterd_volinfo_t *volinfo = NULL; glusterd_brickinfo_t *brickinfo = NULL; + glusterd_snap_t *snap = NULL; GF_ASSERT (this); priv = this->private; GF_ASSERT (priv); + /* Resolve bricks of volumes */ list_for_each_entry (volinfo, &priv->volumes, vol_list) { list_for_each_entry (brickinfo, &volinfo->bricks, brick_list) { ret = glusterd_resolve_brick (brickinfo); @@ -3643,9 +3932,20 @@ glusterd_resolve_all_bricks (xlator_t *this) } } -out: - gf_log ("", GF_LOG_DEBUG, "Returning with %d", ret); + /* Resolve bricks of snapshot volumes */ + list_for_each_entry (snap, &priv->snapshots, snap_list) { + ret = glusterd_resolve_snap_bricks (this, snap); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, + "resolving the snap bricks" + " failed for snap: %s", + snap->snapname); + goto out; + } + } +out: + gf_log (this->name, GF_LOG_TRACE, "Returning with %d", ret); return ret; } @@ -3680,6 +3980,20 @@ glusterd_restore () if (ret) goto out; + ret = glusterd_snap_cleanup (this); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, "Failed to perform " + "a cleanup of the snapshots"); + goto out; + } + + ret = glusterd_recreate_all_snap_brick_mounts (this); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, "Failed to recreate " + "all snap brick mounts"); + goto out; + } + out: gf_log ("", GF_LOG_DEBUG, "Returning %d", ret); return ret; @@ -3722,7 +4036,7 @@ glusterd_store_retrieve_quota_version (glusterd_volinfo_t *volinfo) } version = strtoul (version_str, &tmp, 10); - if (version < 0) { + if ((errno == ERANGE) || (errno == EINVAL)) { gf_log (this->name, GF_LOG_DEBUG, "Invalid version number"); goto out; } diff --git a/xlators/mgmt/glusterd/src/glusterd-store.h b/xlators/mgmt/glusterd/src/glusterd-store.h index 64c073a8a..63d510cbf 100644 --- a/xlators/mgmt/glusterd/src/glusterd-store.h +++ b/xlators/mgmt/glusterd/src/glusterd-store.h @@ -48,7 +48,7 @@ typedef enum glusterd_store_ver_ac_{ #define GLUSTERD_STORE_KEY_VOL_VERSION "version" #define GLUSTERD_STORE_KEY_VOL_TRANSPORT "transport-type" #define GLUSTERD_STORE_KEY_VOL_ID "volume-id" -#define GLUSTERD_STORE_KEY_VOL_IS_RESTORED "is-volume-restored" +#define GLUSTERD_STORE_KEY_VOL_RESTORED_SNAP "restored_from_snap" #define GLUSTERD_STORE_KEY_RB_STATUS "rb_status" #define GLUSTERD_STORE_KEY_RB_SRC_BRICK "rb_src" #define GLUSTERD_STORE_KEY_RB_DST_BRICK "rb_dst" @@ -168,7 +168,6 @@ int32_t glusterd_store_snap (glusterd_snap_t *snap); int32_t -glusterd_store_update_missed_snaps (dict_t *dict, - int32_t missed_snap_count); +glusterd_store_update_missed_snaps (); #endif diff --git a/xlators/mgmt/glusterd/src/glusterd-utils.c b/xlators/mgmt/glusterd/src/glusterd-utils.c index 8e96be91b..7883a98bf 100644 --- a/xlators/mgmt/glusterd/src/glusterd-utils.c +++ b/xlators/mgmt/glusterd/src/glusterd-utils.c @@ -13,6 +13,12 @@ #endif #include <inttypes.h> +#if !defined(__NetBSD__) && !defined(GF_DARWIN_HOST_OS) +#include <mntent.h> +#else +#include "mntent_compat.h" +#endif + #include "globals.h" #include "glusterfs.h" #include "compat.h" @@ -44,7 +50,6 @@ #include <inttypes.h> #include <signal.h> #include <sys/types.h> -#include <net/if.h> #include <sys/ioctl.h> #include <sys/socket.h> #include <rpc/pmap_clnt.h> @@ -56,11 +61,6 @@ #include <lvm2app.h> #endif - -#ifdef GF_LINUX_HOST_OS -#include <mntent.h> -#endif - #ifdef GF_SOLARIS_HOST_OS #include <sys/sockio.h> #endif @@ -709,7 +709,7 @@ glusterd_snap_volinfo_restore (dict_t *rsp_dict, /* Adding missed delete to the dict */ ret = glusterd_add_missed_snaps_to_dict (rsp_dict, - snap_volinfo->volname, + snap_volinfo, brickinfo, brick_count + 1, GF_SNAP_OPTION_TYPE_RESTORE); @@ -2241,12 +2241,17 @@ out: return ret; } +/* The prefix represents the type of volume to be added. + * It will be "volume" for normal volumes, and snap# like + * snap1, snap2, for snapshot volumes + */ int32_t glusterd_add_volume_to_dict (glusterd_volinfo_t *volinfo, - dict_t *dict, int32_t count) + dict_t *dict, int32_t count, + char *prefix) { int32_t ret = -1; - char prefix[512] = {0,}; + char pfx[512] = {0,}; char key[512] = {0,}; glusterd_brickinfo_t *brickinfo = NULL; int32_t i = 1; @@ -2263,89 +2268,101 @@ glusterd_add_volume_to_dict (glusterd_volinfo_t *volinfo, GF_ASSERT (this); GF_ASSERT (dict); GF_ASSERT (volinfo); + GF_ASSERT (prefix); - snprintf (key, sizeof (key), "volume%d.name", count); + snprintf (key, sizeof (key), "%s%d.name", prefix, count); ret = dict_set_str (dict, key, volinfo->volname); if (ret) goto out; memset (key, 0, sizeof (key)); - snprintf (key, sizeof (key), "volume%d.type", count); + snprintf (key, sizeof (key), "%s%d.type", prefix, count); ret = dict_set_int32 (dict, key, volinfo->type); if (ret) goto out; - snprintf (key, sizeof (key), "volume%d.is_volume_restored", count); - ret = dict_set_int32 (dict, key, volinfo->is_volume_restored); - if (ret) { - gf_log (THIS->name, GF_LOG_ERROR, "Failed to set " - "is_volume_restored option for %s volume", - volinfo->volname); + snprintf (key, sizeof (key), "volume%d.restored_from_snap", count); + ret = dict_set_dynstr_with_alloc + (dict, key, + uuid_utoa (volinfo->restored_from_snap)); + if (ret) goto out; + + if (strlen (volinfo->parent_volname) > 0) { + snprintf (key, sizeof (key), "%s%d.parent_volname", + prefix, count); + ret = dict_set_dynstr_with_alloc (dict, key, + volinfo->parent_volname); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, + "Failed to set parent_volname for %s", + volinfo->volname); + goto out; + } } memset (key, 0, sizeof (key)); - snprintf (key, sizeof (key), "volume%d.brick_count", count); + snprintf (key, sizeof (key), "%s%d.brick_count", prefix, count); ret = dict_set_int32 (dict, key, volinfo->brick_count); if (ret) goto out; memset (key, 0, sizeof (key)); - snprintf (key, sizeof (key), "volume%d.version", count); + snprintf (key, sizeof (key), "%s%d.version", prefix, count); ret = dict_set_int32 (dict, key, volinfo->version); if (ret) goto out; memset (key, 0, sizeof (key)); - snprintf (key, sizeof (key), "volume%d.status", count); + snprintf (key, sizeof (key), "%s%d.status", prefix, count); ret = dict_set_int32 (dict, key, volinfo->status); if (ret) goto out; memset (key, 0, sizeof (key)); - snprintf (key, sizeof (key), "volume%d.sub_count", count); + snprintf (key, sizeof (key), "%s%d.sub_count", prefix, count); ret = dict_set_int32 (dict, key, volinfo->sub_count); if (ret) goto out; memset (key, 0, sizeof (key)); - snprintf (key, sizeof (key), "volume%d.stripe_count", count); + snprintf (key, sizeof (key), "%s%d.stripe_count", prefix, count); ret = dict_set_int32 (dict, key, volinfo->stripe_count); if (ret) goto out; memset (key, 0, sizeof (key)); - snprintf (key, sizeof (key), "volume%d.replica_count", count); + snprintf (key, sizeof (key), "%s%d.replica_count", prefix, count); ret = dict_set_int32 (dict, key, volinfo->replica_count); if (ret) goto out; memset (key, 0, sizeof (key)); - snprintf (key, sizeof (key), "volume%d.dist_count", count); + snprintf (key, sizeof (key), "%s%d.dist_count", prefix, count); ret = dict_set_int32 (dict, key, volinfo->dist_leaf_count); if (ret) goto out; memset (key, 0, sizeof (key)); - snprintf (key, sizeof (key), "volume%d.ckusm", count); + snprintf (key, sizeof (key), "%s%d.ckusm", prefix, count); ret = dict_set_int64 (dict, key, volinfo->cksum); if (ret) goto out; memset (key, 0, sizeof (key)); - snprintf (key, sizeof (key), "volume%d.transport_type", count); + snprintf (key, sizeof (key), "%s%d.transport_type", prefix, count); ret = dict_set_uint32 (dict, key, volinfo->transport_type); if (ret) goto out; - snprintf (key, sizeof (key), "volume%d.is_snap_volume", count); + snprintf (key, sizeof (key), "%s%d.is_snap_volume", prefix, count); ret = dict_set_uint32 (dict, key, volinfo->is_snap_volume); if (ret) { gf_log (THIS->name, GF_LOG_ERROR, "Unable to set %s", key); goto out; } - snprintf (key, sizeof (key), "volume%d.snap-max-hard-limit", count); + snprintf (key, sizeof (key), "%s%d.snap-max-hard-limit", prefix, count); ret = dict_set_uint64 (dict, key, volinfo->snap_max_hard_limit); if (ret) { gf_log (THIS->name, GF_LOG_ERROR, "Unable to set %s", key); @@ -2358,14 +2375,14 @@ glusterd_add_volume_to_dict (glusterd_volinfo_t *volinfo, goto out; } memset (key, 0, sizeof (key)); - snprintf (key, sizeof (key), "volume%d.volume_id", count); + snprintf (key, sizeof (key), "%s%d.volume_id", prefix, count); ret = dict_set_dynstr (dict, key, volume_id_str); if (ret) goto out; volume_id_str = NULL; memset (key, 0, sizeof (key)); - snprintf (key, sizeof (key), "volume%d.username", count); + snprintf (key, sizeof (key), "%s%d.username", prefix, count); str = glusterd_auth_get_username (volinfo); if (str) { ret = dict_set_dynstr (dict, key, gf_strdup (str)); @@ -2374,7 +2391,7 @@ glusterd_add_volume_to_dict (glusterd_volinfo_t *volinfo, } memset (key, 0, sizeof (key)); - snprintf (key, sizeof (key), "volume%d.password", count); + snprintf (key, sizeof (key), "%s%d.password", prefix, count); str = glusterd_auth_get_password (volinfo); if (str) { ret = dict_set_dynstr (dict, key, gf_strdup (str)); @@ -2383,7 +2400,7 @@ glusterd_add_volume_to_dict (glusterd_volinfo_t *volinfo, } memset (key, 0, sizeof (key)); - snprintf (key, 256, "volume%d.rebalance", count); + snprintf (key, 256, "%s%d.rebalance", prefix, count); ret = dict_set_int32 (dict, key, volinfo->rebal.defrag_cmd); if (ret) goto out; @@ -2395,22 +2412,22 @@ glusterd_add_volume_to_dict (glusterd_volinfo_t *volinfo, goto out; } memset (key, 0, sizeof (key)); - snprintf (key, 256, "volume%d.rebalance-id", count); + snprintf (key, 256, "%s%d.rebalance-id", prefix, count); ret = dict_set_dynstr (dict, key, rebalance_id_str); if (ret) goto out; rebalance_id_str = NULL; memset (key, 0, sizeof (key)); - snprintf (key, sizeof (key), "volume%d.rebalance-op", count); + snprintf (key, sizeof (key), "%s%d.rebalance-op", prefix, count); ret = dict_set_uint32 (dict, key, volinfo->rebal.op); if (ret) goto out; if (volinfo->rebal.dict) { - snprintf (prefix, sizeof (prefix), "volume%d", count); + snprintf (pfx, sizeof (pfx), "%s%d", prefix, count); ctx.dict = dict; - ctx.prefix = prefix; + ctx.prefix = pfx; ctx.opt_count = 1; ctx.key_name = "rebal-dict-key"; ctx.val_name = "rebal-dict-value"; @@ -2425,7 +2442,7 @@ glusterd_add_volume_to_dict (glusterd_volinfo_t *volinfo, } memset (key, 0, sizeof (key)); - snprintf (key, 256, "volume%d."GLUSTERD_STORE_KEY_RB_STATUS, count); + snprintf (key, 256, "%s%d."GLUSTERD_STORE_KEY_RB_STATUS, prefix, count); ret = dict_set_int32 (dict, key, volinfo->rep_brick.rb_status); if (ret) goto out; @@ -2433,8 +2450,8 @@ glusterd_add_volume_to_dict (glusterd_volinfo_t *volinfo, if (volinfo->rep_brick.rb_status > GF_RB_STATUS_NONE) { memset (key, 0, sizeof (key)); - snprintf (key, 256, "volume%d."GLUSTERD_STORE_KEY_RB_SRC_BRICK, - count); + snprintf (key, 256, "%s%d."GLUSTERD_STORE_KEY_RB_SRC_BRICK, + prefix, count); gf_asprintf (&src_brick, "%s:%s", volinfo->rep_brick.src_brick->hostname, volinfo->rep_brick.src_brick->path); @@ -2443,8 +2460,8 @@ glusterd_add_volume_to_dict (glusterd_volinfo_t *volinfo, goto out; memset (key, 0, sizeof (key)); - snprintf (key, 256, "volume%d."GLUSTERD_STORE_KEY_RB_DST_BRICK, - count); + snprintf (key, 256, "%s%d."GLUSTERD_STORE_KEY_RB_DST_BRICK, + prefix, count); gf_asprintf (&dst_brick, "%s:%s", volinfo->rep_brick.dst_brick->hostname, volinfo->rep_brick.dst_brick->path); @@ -2459,16 +2476,16 @@ glusterd_add_volume_to_dict (glusterd_volinfo_t *volinfo, } memset (key, 0, sizeof (key)); - snprintf (key, sizeof (key), "volume%d.rb_id", count); + snprintf (key, sizeof (key), "%s%d.rb_id", prefix, count); ret = dict_set_dynstr (dict, key, rb_id_str); if (ret) goto out; rb_id_str = NULL; } - snprintf (prefix, sizeof (prefix), "volume%d", count); + snprintf (pfx, sizeof (pfx), "%s%d", prefix, count); ctx.dict = dict; - ctx.prefix = prefix; + ctx.prefix = pfx; ctx.opt_count = 1; ctx.key_name = "key"; ctx.val_name = "value"; @@ -2477,13 +2494,13 @@ glusterd_add_volume_to_dict (glusterd_volinfo_t *volinfo, dict_foreach (volinfo->dict, _add_dict_to_prdict, &ctx); ctx.opt_count--; memset (key, 0, sizeof (key)); - snprintf (key, sizeof (key), "volume%d.opt-count", count); + snprintf (key, sizeof (key), "%s%d.opt-count", prefix, count); ret = dict_set_int32 (dict, key, ctx.opt_count); if (ret) goto out; ctx.dict = dict; - ctx.prefix = prefix; + ctx.prefix = pfx; ctx.opt_count = 1; ctx.key_name = "slave-num"; ctx.val_name = "slave-val"; @@ -2493,42 +2510,42 @@ glusterd_add_volume_to_dict (glusterd_volinfo_t *volinfo, ctx.opt_count--; memset (key, 0, sizeof (key)); - snprintf (key, sizeof (key), "volume%d.gsync-count", count); + snprintf (key, sizeof (key), "%s%d.gsync-count", prefix, count); ret = dict_set_int32 (dict, key, ctx.opt_count); if (ret) goto out; list_for_each_entry (brickinfo, &volinfo->bricks, brick_list) { memset (key, 0, sizeof (key)); - snprintf (key, sizeof (key), "volume%d.brick%d.hostname", - count, i); + snprintf (key, sizeof (key), "%s%d.brick%d.hostname", + prefix, count, i); ret = dict_set_str (dict, key, brickinfo->hostname); if (ret) goto out; memset (key, 0, sizeof (key)); - snprintf (key, sizeof (key), "volume%d.brick%d.path", - count, i); + snprintf (key, sizeof (key), "%s%d.brick%d.path", + prefix, count, i); ret = dict_set_str (dict, key, brickinfo->path); if (ret) goto out; memset (key, 0, sizeof (key)); - snprintf (key, sizeof (key), "volume%d.brick%d.decommissioned", - count, i); + snprintf (key, sizeof (key), "%s%d.brick%d.decommissioned", + prefix, count, i); ret = dict_set_int32 (dict, key, brickinfo->decommissioned); if (ret) goto out; memset (key, 0, sizeof (key)); - snprintf (key, sizeof (key), "volume%d.brick%d.brick_id", - count, i); + snprintf (key, sizeof (key), "%s%d.brick%d.brick_id", + prefix, count, i); ret = dict_set_str (dict, key, brickinfo->brick_id); if (ret) goto out; - snprintf (key, sizeof (key), "volume%d.brick%d.snap_status", - count, i); + snprintf (key, sizeof (key), "%s%d.brick%d.snap_status", + prefix, count, i); ret = dict_set_int32 (dict, key, brickinfo->snap_status); if (ret) { gf_log (this->name, GF_LOG_ERROR, @@ -2538,8 +2555,8 @@ glusterd_add_volume_to_dict (glusterd_volinfo_t *volinfo, goto out; } - snprintf (key, sizeof (key), "volume%d.brick%d.device_path", - count, i); + snprintf (key, sizeof (key), "%s%d.brick%d.device_path", + prefix, count, i); ret = dict_set_str (dict, key, brickinfo->device_path); if (ret) { gf_log (this->name, GF_LOG_ERROR, @@ -2556,19 +2573,19 @@ glusterd_add_volume_to_dict (glusterd_volinfo_t *volinfo, * in the cluster */ memset (key, 0, sizeof (key)); - snprintf (key, sizeof (key), "volume%d.op-version", count); + snprintf (key, sizeof (key), "%s%d.op-version", prefix, count); ret = dict_set_int32 (dict, key, volinfo->op_version); if (ret) goto out; memset (key, 0, sizeof (key)); - snprintf (key, sizeof (key), "volume%d.client-op-version", count); + snprintf (key, sizeof (key), "%s%d.client-op-version", prefix, count); ret = dict_set_int32 (dict, key, volinfo->client_op_version); if (ret) goto out; /*Add volume Capability (BD Xlator) to dict*/ memset (key, 0 ,sizeof (key)); - snprintf (key, sizeof (key), "volume%d.caps", count); + snprintf (key, sizeof (key), "%s%d.caps", prefix, count); ret = dict_set_int32 (dict, key, volinfo->caps); out: @@ -2576,14 +2593,17 @@ out: GF_FREE (rebalance_id_str); GF_FREE (rb_id_str); - gf_log ("", GF_LOG_DEBUG, "Returning with %d", ret); - + gf_log (this->name, GF_LOG_DEBUG, "Returning with %d", ret); return ret; } +/* The prefix represents the type of volume to be added. + * It will be "volume" for normal volumes, and snap# like + * snap1, snap2, for snapshot volumes + */ int glusterd_vol_add_quota_conf_to_dict (glusterd_volinfo_t *volinfo, dict_t* load, - int vol_idx) + int vol_idx, char *prefix) { int fd = -1; char *gfid_str = NULL; @@ -2595,6 +2615,7 @@ glusterd_vol_add_quota_conf_to_dict (glusterd_volinfo_t *volinfo, dict_t* load, this = THIS; GF_ASSERT (this); + GF_ASSERT (prefix); ret = glusterd_store_create_quota_conf_sh_on_absence (volinfo); if (ret) @@ -2631,8 +2652,8 @@ glusterd_vol_add_quota_conf_to_dict (glusterd_volinfo_t *volinfo, dict_t* load, goto out; } - snprintf (key, sizeof(key)-1, "volume%d.gfid%d", vol_idx, - gfid_idx); + snprintf (key, sizeof(key)-1, "%s%d.gfid%d", prefix, + vol_idx, gfid_idx); key[sizeof(key)-1] = '\0'; ret = dict_set_dynstr (load, key, gfid_str); if (ret) { @@ -2642,19 +2663,19 @@ glusterd_vol_add_quota_conf_to_dict (glusterd_volinfo_t *volinfo, dict_t* load, gfid_str = NULL; } - snprintf (key, sizeof(key)-1, "volume%d.gfid-count", vol_idx); + snprintf (key, sizeof(key)-1, "%s%d.gfid-count", prefix, vol_idx); key[sizeof(key)-1] = '\0'; ret = dict_set_int32 (load, key, gfid_idx); if (ret) goto out; - snprintf (key, sizeof(key)-1, "volume%d.quota-cksum", vol_idx); + snprintf (key, sizeof(key)-1, "%s%d.quota-cksum", prefix, vol_idx); key[sizeof(key)-1] = '\0'; ret = dict_set_uint32 (load, key, volinfo->quota_conf_cksum); if (ret) goto out; - snprintf (key, sizeof(key)-1, "volume%d.quota-version", vol_idx); + snprintf (key, sizeof(key)-1, "%s%d.quota-version", prefix, vol_idx); key[sizeof(key)-1] = '\0'; ret = dict_set_uint32 (load, key, volinfo->quota_conf_version); if (ret) @@ -2669,7 +2690,237 @@ out: } int32_t -glusterd_build_volume_dict (dict_t **vols) +glusterd_add_missed_snaps_to_export_dict (dict_t *peer_data) +{ + char name_buf[PATH_MAX] = ""; + char value[PATH_MAX] = ""; + int32_t missed_snap_count = 0; + int32_t ret = -1; + glusterd_conf_t *priv = NULL; + glusterd_missed_snap_info *missed_snapinfo = NULL; + glusterd_snap_op_t *snap_opinfo = NULL; + xlator_t *this = NULL; + + this = THIS; + GF_ASSERT (this); + GF_ASSERT (peer_data); + + priv = this->private; + GF_ASSERT (priv); + + /* Add the missed_entries in the dict */ + list_for_each_entry (missed_snapinfo, &priv->missed_snaps_list, + missed_snaps) { + list_for_each_entry (snap_opinfo, + &missed_snapinfo->snap_ops, + snap_ops_list) { + snprintf (name_buf, sizeof(name_buf), + "missed_snaps_%d", missed_snap_count); + snprintf (value, sizeof(value), "%s:%s=%s:%d:%s:%d:%d", + missed_snapinfo->node_uuid, + missed_snapinfo->snap_uuid, + snap_opinfo->snap_vol_id, + snap_opinfo->brick_num, + snap_opinfo->brick_path, + snap_opinfo->op, + snap_opinfo->status); + + ret = dict_set_dynstr_with_alloc (peer_data, name_buf, + value); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, + "Unable to set %s", + name_buf); + goto out; + } + missed_snap_count++; + } + } + + ret = dict_set_int32 (peer_data, "missed_snap_count", + missed_snap_count); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, + "Unable to set missed_snap_count"); + goto out; + } + +out: + gf_log (this->name, GF_LOG_TRACE, "Returning %d", ret); + return ret; +} + +int32_t +glusterd_add_snap_to_dict (glusterd_snap_t *snap, dict_t *peer_data, + int32_t snap_count) +{ + char buf[NAME_MAX] = ""; + char prefix[NAME_MAX] = ""; + int32_t ret = -1; + int32_t volcount = 0; + glusterd_volinfo_t *volinfo = NULL; + glusterd_brickinfo_t *brickinfo = NULL; + gf_boolean_t host_bricks = _gf_false; + xlator_t *this = NULL; + + this = THIS; + GF_ASSERT (this); + GF_ASSERT (snap); + GF_ASSERT (peer_data); + + snprintf (prefix, sizeof(prefix), "snap%d", snap_count); + + list_for_each_entry (volinfo, &snap->volumes, vol_list) { + volcount++; + ret = glusterd_add_volume_to_dict (volinfo, peer_data, + volcount, prefix); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, + "Failed to add snap:%s volume:%s " + "to peer_data dict for handshake", + snap->snapname, volinfo->volname); + goto out; + } + + ret = glusterd_vol_add_quota_conf_to_dict (volinfo, peer_data, + volcount, prefix); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, + "Failed to add quota conf for " + "snap:%s volume:%s to peer_data " + "dict for handshake", snap->snapname, + volinfo->volname); + goto out; + } + + list_for_each_entry (brickinfo, &volinfo->bricks, brick_list) { + if (!uuid_compare (brickinfo->uuid, MY_UUID)) { + host_bricks = _gf_true; + break; + } + } + } + + snprintf (buf, sizeof(buf), "%s.host_bricks", prefix); + ret = dict_set_int8 (peer_data, buf, (int8_t) host_bricks); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, + "Unable to set host_bricks for snap %s", + snap->snapname); + goto out; + } + + snprintf (buf, sizeof(buf), "%s.volcount", prefix); + ret = dict_set_int32 (peer_data, buf, volcount); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, + "Unable to set volcount for snap %s", + snap->snapname); + goto out; + } + + snprintf (buf, sizeof(buf), "%s.snapname", prefix); + ret = dict_set_dynstr_with_alloc (peer_data, buf, snap->snapname); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, + "Unable to set snapname for snap %s", + snap->snapname); + goto out; + } + + snprintf (buf, sizeof(buf), "%s.snap_id", prefix); + ret = dict_set_dynstr_with_alloc (peer_data, buf, + uuid_utoa (snap->snap_id)); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, + "Unable to set snap_id for snap %s", + snap->snapname); + goto out; + } + + if (snap->description) { + snprintf (buf, sizeof(buf), "%s.snapid", prefix); + ret = dict_set_dynstr_with_alloc (peer_data, buf, + snap->description); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, + "Unable to set description for snap %s", + snap->snapname); + goto out; + } + } + + snprintf (buf, sizeof(buf), "%s.time_stamp", prefix); + ret = dict_set_int64 (peer_data, buf, (int64_t)snap->time_stamp); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, + "Unable to set time_stamp for snap %s", + snap->snapname); + goto out; + } + + snprintf (buf, sizeof(buf), "%s.snap_restored", prefix); + ret = dict_set_int8 (peer_data, buf, snap->snap_restored); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, + "Unable to set snap_restored for snap %s", + snap->snapname); + goto out; + } + + snprintf (buf, sizeof(buf), "%s.snap_status", prefix); + ret = dict_set_int32 (peer_data, buf, snap->snap_status); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, + "Unable to set snap_status for snap %s", + snap->snapname); + goto out; + } +out: + gf_log (this->name, GF_LOG_TRACE, "Returning %d", ret); + return ret; +} + +int32_t +glusterd_add_snapshots_to_export_dict (dict_t *peer_data) +{ + int32_t snap_count = 0; + int32_t ret = -1; + glusterd_conf_t *priv = NULL; + glusterd_snap_t *snap = NULL; + xlator_t *this = NULL; + + this = THIS; + GF_ASSERT (this); + priv = this->private; + GF_ASSERT (priv); + GF_ASSERT (peer_data); + + list_for_each_entry (snap, &priv->snapshots, snap_list) { + snap_count++; + ret = glusterd_add_snap_to_dict (snap, peer_data, snap_count); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, + "Failed to add snap(%s) to the " + " peer_data dict for handshake", + snap->snapname); + goto out; + } + } + + ret = dict_set_int32 (peer_data, "snap_count", snap_count); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, "Failed to set snap_count"); + goto out; + } + +out: + gf_log (this->name, GF_LOG_TRACE, "Returning %d", ret); + return ret; +} + +int32_t +glusterd_add_volumes_to_export_dict (dict_t **peer_data) { int32_t ret = -1; dict_t *dict = NULL; @@ -2677,27 +2928,31 @@ glusterd_build_volume_dict (dict_t **vols) glusterd_volinfo_t *volinfo = NULL; int32_t count = 0; glusterd_dict_ctx_t ctx = {0}; + xlator_t *this = NULL; - priv = THIS->private; + this = THIS; + GF_ASSERT (this); + priv = this->private; + GF_ASSERT (priv); dict = dict_new (); - if (!dict) goto out; list_for_each_entry (volinfo, &priv->volumes, vol_list) { count++; - ret = glusterd_add_volume_to_dict (volinfo, dict, count); + ret = glusterd_add_volume_to_dict (volinfo, dict, count, + "volume"); if (ret) goto out; if (!glusterd_is_volume_quota_enabled (volinfo)) continue; - ret = glusterd_vol_add_quota_conf_to_dict (volinfo, dict, count); + ret = glusterd_vol_add_quota_conf_to_dict (volinfo, dict, + count, "volume"); if (ret) goto out; } - ret = dict_set_int32 (dict, "count", count); if (ret) goto out; @@ -2713,18 +2968,18 @@ glusterd_build_volume_dict (dict_t **vols) if (ret) goto out; - *vols = dict; + *peer_data = dict; out: - gf_log ("", GF_LOG_DEBUG, "Returning with %d", ret); if (ret) dict_unref (dict); + gf_log (this->name, GF_LOG_TRACE, "Returning %d", ret); return ret; } int32_t -glusterd_compare_friend_volume (dict_t *vols, int32_t count, int32_t *status, - char *hostname) +glusterd_compare_friend_volume (dict_t *peer_data, int32_t count, + int32_t *status, char *hostname) { int32_t ret = -1; @@ -2737,14 +2992,14 @@ glusterd_compare_friend_volume (dict_t *vols, int32_t count, int32_t *status, int32_t version = 0; xlator_t *this = NULL; - GF_ASSERT (vols); + GF_ASSERT (peer_data); GF_ASSERT (status); this = THIS; GF_ASSERT (this); snprintf (key, sizeof (key), "volume%d.name", count); - ret = dict_get_str (vols, key, &volname); + ret = dict_get_str (peer_data, key, &volname); if (ret) goto out; @@ -2758,7 +3013,7 @@ glusterd_compare_friend_volume (dict_t *vols, int32_t count, int32_t *status, memset (key, 0, sizeof (key)); snprintf (key, sizeof (key), "volume%d.version", count); - ret = dict_get_int32 (vols, key, &version); + ret = dict_get_int32 (peer_data, key, &version); if (ret) goto out; @@ -2779,7 +3034,7 @@ glusterd_compare_friend_volume (dict_t *vols, int32_t count, int32_t *status, // memset (key, 0, sizeof (key)); snprintf (key, sizeof (key), "volume%d.ckusm", count); - ret = dict_get_uint32 (vols, key, &cksum); + ret = dict_get_uint32 (peer_data, key, &cksum); if (ret) goto out; @@ -2794,7 +3049,7 @@ glusterd_compare_friend_volume (dict_t *vols, int32_t count, int32_t *status, memset (key, 0, sizeof (key)); snprintf (key, sizeof (key), "volume%d.quota-version", count); - ret = dict_get_uint32 (vols, key, "a_version); + ret = dict_get_uint32 (peer_data, key, "a_version); if (ret) { gf_log (this->name, GF_LOG_DEBUG, "quota-version key absent for" " volume %s in peer %s's response", volinfo->volname, @@ -2822,7 +3077,7 @@ glusterd_compare_friend_volume (dict_t *vols, int32_t count, int32_t *status, // memset (key, 0, sizeof (key)); snprintf (key, sizeof (key), "volume%d.quota-cksum", count); - ret = dict_get_uint32 (vols, key, "a_cksum); + ret = dict_get_uint32 (peer_data, key, "a_cksum); if (ret) { gf_log (this->name, GF_LOG_DEBUG, "quota checksum absent for " "volume %s in peer %s's response", volinfo->volname, @@ -2850,7 +3105,7 @@ out: } static int32_t -import_prdict_dict (dict_t *vols, dict_t *dst_dict, char *key_prefix, +import_prdict_dict (dict_t *peer_data, dict_t *dst_dict, char *key_prefix, char *value_prefix, int opt_count, char *prefix) { char key[512] = {0,}; @@ -2865,7 +3120,7 @@ import_prdict_dict (dict_t *vols, dict_t *dst_dict, char *key_prefix, memset (key, 0, sizeof (key)); snprintf (key, sizeof (key), "%s.%s%d", prefix, key_prefix, i); - ret = dict_get_str (vols, key, &opt_key); + ret = dict_get_str (peer_data, key, &opt_key); if (ret) { snprintf (msg, sizeof (msg), "Volume dict key not " "specified"); @@ -2875,7 +3130,7 @@ import_prdict_dict (dict_t *vols, dict_t *dst_dict, char *key_prefix, memset (key, 0, sizeof (key)); snprintf (key, sizeof (key), "%s.%s%d", prefix, value_prefix, i); - ret = dict_get_str (vols, key, &opt_val); + ret = dict_get_str (peer_data, key, &opt_val); if (ret) { snprintf (msg, sizeof (msg), "Volume dict value not " "specified"); @@ -3163,7 +3418,7 @@ out: } int32_t -glusterd_import_friend_volume_opts (dict_t *vols, int count, +glusterd_import_friend_volume_opts (dict_t *peer_data, int count, glusterd_volinfo_t *volinfo) { char key[512] = {0,}; @@ -3172,9 +3427,12 @@ glusterd_import_friend_volume_opts (dict_t *vols, int count, char msg[2048] = {0}; char volume_prefix[1024] = {0}; + GF_ASSERT (peer_data); + GF_ASSERT (volinfo); + memset (key, 0, sizeof (key)); snprintf (key, sizeof (key), "volume%d.opt-count", count); - ret = dict_get_int32 (vols, key, &opt_count); + ret = dict_get_int32 (peer_data, key, &opt_count); if (ret) { snprintf (msg, sizeof (msg), "Volume option count not " "specified for %s", volinfo->volname); @@ -3182,7 +3440,7 @@ glusterd_import_friend_volume_opts (dict_t *vols, int count, } snprintf (volume_prefix, sizeof (volume_prefix), "volume%d", count); - ret = import_prdict_dict (vols, volinfo->dict, "key", "value", + ret = import_prdict_dict (peer_data, volinfo->dict, "key", "value", opt_count, volume_prefix); if (ret) { snprintf (msg, sizeof (msg), "Unable to import options dict " @@ -3192,14 +3450,14 @@ glusterd_import_friend_volume_opts (dict_t *vols, int count, memset (key, 0, sizeof (key)); snprintf (key, sizeof (key), "volume%d.gsync-count", count); - ret = dict_get_int32 (vols, key, &opt_count); + ret = dict_get_int32 (peer_data, key, &opt_count); if (ret) { snprintf (msg, sizeof (msg), "Gsync count not " "specified for %s", volinfo->volname); goto out; } - ret = import_prdict_dict (vols, volinfo->gsync_slaves, "slave-num", + ret = import_prdict_dict (peer_data, volinfo->gsync_slaves, "slave-num", "slave-val", opt_count, volume_prefix); if (ret) { snprintf (msg, sizeof (msg), "Unable to import gsync sessions " @@ -3214,10 +3472,15 @@ out: return ret; } +/* The prefix represents the type of volume to be added. + * It will be "volume" for normal volumes, and snap# like + * snap1, snap2, for snapshot volumes + */ int32_t -glusterd_import_new_brick (dict_t *vols, int32_t vol_count, +glusterd_import_new_brick (dict_t *peer_data, int32_t vol_count, int32_t brick_count, - glusterd_brickinfo_t **brickinfo) + glusterd_brickinfo_t **brickinfo, + char *prefix) { char key[512] = {0,}; int ret = -1; @@ -3230,53 +3493,54 @@ glusterd_import_new_brick (dict_t *vols, int32_t vol_count, glusterd_brickinfo_t *new_brickinfo = NULL; char msg[2048] = {0}; - GF_ASSERT (vols); + GF_ASSERT (peer_data); GF_ASSERT (vol_count >= 0); GF_ASSERT (brickinfo); + GF_ASSERT (prefix); memset (key, 0, sizeof (key)); - snprintf (key, sizeof (key), "volume%d.brick%d.hostname", - vol_count, brick_count); - ret = dict_get_str (vols, key, &hostname); + snprintf (key, sizeof (key), "%s%d.brick%d.hostname", + prefix, vol_count, brick_count); + ret = dict_get_str (peer_data, key, &hostname); if (ret) { snprintf (msg, sizeof (msg), "%s missing in payload", key); goto out; } memset (key, 0, sizeof (key)); - snprintf (key, sizeof (key), "volume%d.brick%d.path", - vol_count, brick_count); - ret = dict_get_str (vols, key, &path); + snprintf (key, sizeof (key), "%s%d.brick%d.path", + prefix, vol_count, brick_count); + ret = dict_get_str (peer_data, key, &path); if (ret) { snprintf (msg, sizeof (msg), "%s missing in payload", key); goto out; } memset (key, 0, sizeof (key)); - snprintf (key, sizeof (key), "volume%d.brick%d.brick_id", - vol_count, brick_count); - ret = dict_get_str (vols, key, &brick_id); + snprintf (key, sizeof (key), "%s%d.brick%d.brick_id", + prefix, vol_count, brick_count); + ret = dict_get_str (peer_data, key, &brick_id); memset (key, 0, sizeof (key)); - snprintf (key, sizeof (key), "volume%d.brick%d.decommissioned", - vol_count, brick_count); - ret = dict_get_int32 (vols, key, &decommissioned); + snprintf (key, sizeof (key), "%s%d.brick%d.decommissioned", + prefix, vol_count, brick_count); + ret = dict_get_int32 (peer_data, key, &decommissioned); if (ret) { /* For backward compatibility */ ret = 0; } - snprintf (key, sizeof (key), "volume%d.brick%d.snap_status", - vol_count, brick_count); - ret = dict_get_int32 (vols, key, &snap_status); + snprintf (key, sizeof (key), "%s%d.brick%d.snap_status", + prefix, vol_count, brick_count); + ret = dict_get_int32 (peer_data, key, &snap_status); if (ret) { snprintf (msg, sizeof (msg), "%s missing in payload", key); goto out; } - snprintf (key, sizeof (key), "volume%d.brick%d.device_path", - vol_count, brick_count); - ret = dict_get_str (vols, key, &snap_device); + snprintf (key, sizeof (key), "%s%d.brick%d.device_path", + prefix, vol_count, brick_count); + ret = dict_get_str (peer_data, key, &snap_device); if (ret) { snprintf (msg, sizeof (msg), "%s missing in payload", key); goto out; @@ -3304,22 +3568,28 @@ out: return ret; } +/* The prefix represents the type of volume to be added. + * It will be "volume" for normal volumes, and snap# like + * snap1, snap2, for snapshot volumes + */ int32_t -glusterd_import_bricks (dict_t *vols, int32_t vol_count, - glusterd_volinfo_t *new_volinfo) +glusterd_import_bricks (dict_t *peer_data, int32_t vol_count, + glusterd_volinfo_t *new_volinfo, char *prefix) { int ret = -1; int brick_count = 1; int brickid = 0; glusterd_brickinfo_t *new_brickinfo = NULL; - GF_ASSERT (vols); + GF_ASSERT (peer_data); GF_ASSERT (vol_count >= 0); GF_ASSERT (new_volinfo); + GF_ASSERT (prefix); while (brick_count <= new_volinfo->brick_count) { - ret = glusterd_import_new_brick (vols, vol_count, brick_count, - &new_brickinfo); + ret = glusterd_import_new_brick (peer_data, vol_count, + brick_count, + &new_brickinfo, prefix); if (ret) goto out; if (new_brickinfo->brick_id[0] == '\0') @@ -3337,9 +3607,14 @@ out: return ret; } +/* The prefix represents the type of volume to be added. + * It will be "volume" for normal volumes, and snap# like + * snap1, snap2, for snapshot volumes + */ static int -glusterd_import_quota_conf (dict_t *vols, int vol_idx, - glusterd_volinfo_t *new_volinfo) +glusterd_import_quota_conf (dict_t *peer_data, int vol_idx, + glusterd_volinfo_t *new_volinfo, + char *prefix) { int gfid_idx = 0; int gfid_count = 0; @@ -3352,6 +3627,8 @@ glusterd_import_quota_conf (dict_t *vols, int vol_idx, this = THIS; GF_ASSERT (this); + GF_ASSERT (peer_data); + GF_ASSERT (prefix); if (!glusterd_is_volume_quota_enabled (new_volinfo)) { (void) glusterd_clean_up_quota_store (new_volinfo); @@ -3368,22 +3645,23 @@ glusterd_import_quota_conf (dict_t *vols, int vol_idx, goto out; } - snprintf (key, sizeof (key)-1, "volume%d.quota-cksum", vol_idx); + snprintf (key, sizeof (key)-1, "%s%d.quota-cksum", prefix, vol_idx); key[sizeof(key)-1] = '\0'; - ret = dict_get_uint32 (vols, key, &new_volinfo->quota_conf_cksum); + ret = dict_get_uint32 (peer_data, key, &new_volinfo->quota_conf_cksum); if (ret) gf_log (this->name, GF_LOG_DEBUG, "Failed to get quota cksum"); - snprintf (key, sizeof (key)-1, "volume%d.quota-version", vol_idx); + snprintf (key, sizeof (key)-1, "%s%d.quota-version", prefix, vol_idx); key[sizeof(key)-1] = '\0'; - ret = dict_get_uint32 (vols, key, &new_volinfo->quota_conf_version); + ret = dict_get_uint32 (peer_data, key, + &new_volinfo->quota_conf_version); if (ret) gf_log (this->name, GF_LOG_DEBUG, "Failed to get quota " "version"); - snprintf (key, sizeof (key)-1, "volume%d.gfid-count", vol_idx); + snprintf (key, sizeof (key)-1, "%s%d.gfid-count", prefix, vol_idx); key[sizeof(key)-1] = '\0'; - ret = dict_get_int32 (vols, key, &gfid_count); + ret = dict_get_int32 (peer_data, key, &gfid_count); if (ret) goto out; @@ -3397,10 +3675,10 @@ glusterd_import_quota_conf (dict_t *vols, int vol_idx, gfid_idx = 0; for (gfid_idx = 0; gfid_idx < gfid_count; gfid_idx++) { - snprintf (key, sizeof (key)-1, "volume%d.gfid%d", - vol_idx, gfid_idx); + snprintf (key, sizeof (key)-1, "%s%d.gfid%d", + prefix, vol_idx, gfid_idx); key[sizeof(key)-1] = '\0'; - ret = dict_get_str (vols, key, &gfid_str); + ret = dict_get_str (peer_data, key, &gfid_str); if (ret) goto out; @@ -3481,15 +3759,22 @@ out: return ret; } +/* The prefix represents the type of volume to be added. + * It will be "volume" for normal volumes, and snap# like + * snap1, snap2, for snapshot volumes + */ int32_t -glusterd_import_volinfo (dict_t *vols, int count, - glusterd_volinfo_t **volinfo) +glusterd_import_volinfo (dict_t *peer_data, int count, + glusterd_volinfo_t **volinfo, + char *prefix) { int ret = -1; char key[256] = {0}; + char *parent_volname = NULL; char *volname = NULL; glusterd_volinfo_t *new_volinfo = NULL; char *volume_id_str = NULL; + char *restored_snap = NULL; char msg[2048] = {0}; char *src_brick = NULL; char *dst_brick = NULL; @@ -3501,50 +3786,49 @@ glusterd_import_volinfo (dict_t *vols, int count, int client_op_version = 0; uint32_t is_snap_volume = 0; - GF_ASSERT (vols); + GF_ASSERT (peer_data); GF_ASSERT (volinfo); + GF_ASSERT (prefix); - snprintf (key, sizeof (key), "volume%d.name", count); - ret = dict_get_str (vols, key, &volname); + snprintf (key, sizeof (key), "%s%d.name", prefix, count); + ret = dict_get_str (peer_data, key, &volname); if (ret) { snprintf (msg, sizeof (msg), "%s missing in payload", key); goto out; } memset (key, 0, sizeof (key)); - snprintf (key, sizeof (key), "volume%d.is_snap_volume", count); - ret = dict_get_uint32 (vols, key, &is_snap_volume); + snprintf (key, sizeof (key), "%s%d.is_snap_volume", prefix, count); + ret = dict_get_uint32 (peer_data, key, &is_snap_volume); if (ret) { snprintf (msg, sizeof (msg), "%s missing in payload for %s", key, volname); goto out; } - if (is_snap_volume == _gf_true) { - gf_log (THIS->name, GF_LOG_DEBUG, - "Not syncing snap volume %s", volname); - ret = 0; - goto out; - } - ret = glusterd_volinfo_new (&new_volinfo); if (ret) goto out; strncpy (new_volinfo->volname, volname, sizeof (new_volinfo->volname)); - memset (key, 0, sizeof (key)); - snprintf (key, sizeof (key), "volume%d.type", count); - ret = dict_get_int32 (vols, key, &new_volinfo->type); + snprintf (key, sizeof (key), "%s%d.type", prefix, count); + ret = dict_get_int32 (peer_data, key, &new_volinfo->type); if (ret) { snprintf (msg, sizeof (msg), "%s missing in payload for %s", key, volname); goto out; } + snprintf (key, sizeof (key), "%s%d.parent_volname", prefix, count); + ret = dict_get_str (peer_data, key, &parent_volname); + if (!ret) + strncpy (new_volinfo->parent_volname, parent_volname, + sizeof(new_volinfo->parent_volname)); + memset (key, 0, sizeof (key)); - snprintf (key, sizeof (key), "volume%d.brick_count", count); - ret = dict_get_int32 (vols, key, &new_volinfo->brick_count); + snprintf (key, sizeof (key), "%s%d.brick_count", prefix, count); + ret = dict_get_int32 (peer_data, key, &new_volinfo->brick_count); if (ret) { snprintf (msg, sizeof (msg), "%s missing in payload for %s", key, volname); @@ -3552,8 +3836,8 @@ glusterd_import_volinfo (dict_t *vols, int count, } memset (key, 0, sizeof (key)); - snprintf (key, sizeof (key), "volume%d.version", count); - ret = dict_get_int32 (vols, key, &new_volinfo->version); + snprintf (key, sizeof (key), "%s%d.version", prefix, count); + ret = dict_get_int32 (peer_data, key, &new_volinfo->version); if (ret) { snprintf (msg, sizeof (msg), "%s missing in payload for %s", key, volname); @@ -3561,8 +3845,8 @@ glusterd_import_volinfo (dict_t *vols, int count, } memset (key, 0, sizeof (key)); - snprintf (key, sizeof (key), "volume%d.status", count); - ret = dict_get_int32 (vols, key, (int32_t *)&new_volinfo->status); + snprintf (key, sizeof (key), "%s%d.status", prefix, count); + ret = dict_get_int32 (peer_data, key, (int32_t *)&new_volinfo->status); if (ret) { snprintf (msg, sizeof (msg), "%s missing in payload for %s", key, volname); @@ -3570,8 +3854,8 @@ glusterd_import_volinfo (dict_t *vols, int count, } memset (key, 0, sizeof (key)); - snprintf (key, sizeof (key), "volume%d.sub_count", count); - ret = dict_get_int32 (vols, key, &new_volinfo->sub_count); + snprintf (key, sizeof (key), "%s%d.sub_count", prefix, count); + ret = dict_get_int32 (peer_data, key, &new_volinfo->sub_count); if (ret) { snprintf (msg, sizeof (msg), "%s missing in payload for %s", key, volname); @@ -3581,8 +3865,8 @@ glusterd_import_volinfo (dict_t *vols, int count, /* not having a 'stripe_count' key is not a error (as peer may be of old version) */ memset (key, 0, sizeof (key)); - snprintf (key, sizeof (key), "volume%d.stripe_count", count); - ret = dict_get_int32 (vols, key, &new_volinfo->stripe_count); + snprintf (key, sizeof (key), "%s%d.stripe_count", prefix, count); + ret = dict_get_int32 (peer_data, key, &new_volinfo->stripe_count); if (ret) gf_log (THIS->name, GF_LOG_INFO, "peer is possibly old version"); @@ -3590,8 +3874,8 @@ glusterd_import_volinfo (dict_t *vols, int count, /* not having a 'replica_count' key is not a error (as peer may be of old version) */ memset (key, 0, sizeof (key)); - snprintf (key, sizeof (key), "volume%d.replica_count", count); - ret = dict_get_int32 (vols, key, &new_volinfo->replica_count); + snprintf (key, sizeof (key), "%s%d.replica_count", prefix, count); + ret = dict_get_int32 (peer_data, key, &new_volinfo->replica_count); if (ret) gf_log (THIS->name, GF_LOG_INFO, "peer is possibly old version"); @@ -3599,16 +3883,16 @@ glusterd_import_volinfo (dict_t *vols, int count, /* not having a 'dist_count' key is not a error (as peer may be of old version) */ memset (key, 0, sizeof (key)); - snprintf (key, sizeof (key), "volume%d.dist_count", count); - ret = dict_get_int32 (vols, key, &new_volinfo->dist_leaf_count); + snprintf (key, sizeof (key), "%s%d.dist_count", prefix, count); + ret = dict_get_int32 (peer_data, key, &new_volinfo->dist_leaf_count); if (ret) gf_log (THIS->name, GF_LOG_INFO, "peer is possibly old version"); new_volinfo->subvol_count = new_volinfo->brick_count/ glusterd_get_dist_leaf_count (new_volinfo); memset (key, 0, sizeof (key)); - snprintf (key, sizeof (key), "volume%d.ckusm", count); - ret = dict_get_uint32 (vols, key, &new_volinfo->cksum); + snprintf (key, sizeof (key), "%s%d.ckusm", prefix, count); + ret = dict_get_uint32 (peer_data, key, &new_volinfo->cksum); if (ret) { snprintf (msg, sizeof (msg), "%s missing in payload for %s", key, volname); @@ -3616,8 +3900,8 @@ glusterd_import_volinfo (dict_t *vols, int count, } memset (key, 0, sizeof (key)); - snprintf (key, sizeof (key), "volume%d.volume_id", count); - ret = dict_get_str (vols, key, &volume_id_str); + snprintf (key, sizeof (key), "%s%d.volume_id", prefix, count); + ret = dict_get_str (peer_data, key, &volume_id_str); if (ret) { snprintf (msg, sizeof (msg), "%s missing in payload for %s", key, volname); @@ -3627,8 +3911,8 @@ glusterd_import_volinfo (dict_t *vols, int count, uuid_parse (volume_id_str, new_volinfo->volume_id); memset (key, 0, sizeof (key)); - snprintf (key, sizeof (key), "volume%d.username", count); - ret = dict_get_str (vols, key, &str); + snprintf (key, sizeof (key), "%s%d.username", prefix, count); + ret = dict_get_str (peer_data, key, &str); if (!ret) { ret = glusterd_auth_set_username (new_volinfo, str); if (ret) @@ -3636,8 +3920,8 @@ glusterd_import_volinfo (dict_t *vols, int count, } memset (key, 0, sizeof (key)); - snprintf (key, sizeof (key), "volume%d.password", count); - ret = dict_get_str (vols, key, &str); + snprintf (key, sizeof (key), "%s%d.password", prefix, count); + ret = dict_get_str (peer_data, key, &str); if (!ret) { ret = glusterd_auth_set_password (new_volinfo, str); if (ret) @@ -3645,8 +3929,8 @@ glusterd_import_volinfo (dict_t *vols, int count, } memset (key, 0, sizeof (key)); - snprintf (key, sizeof (key), "volume%d.transport_type", count); - ret = dict_get_uint32 (vols, key, &new_volinfo->transport_type); + snprintf (key, sizeof (key), "%s%d.transport_type", prefix, count); + ret = dict_get_uint32 (peer_data, key, &new_volinfo->transport_type); if (ret) { snprintf (msg, sizeof (msg), "%s missing in payload for %s", key, volname); @@ -3655,17 +3939,19 @@ glusterd_import_volinfo (dict_t *vols, int count, new_volinfo->is_snap_volume = is_snap_volume; - snprintf (key, sizeof (key), "volume%d.is_volume_restored", count); - ret = dict_get_uint32 (vols, key, &new_volinfo->is_volume_restored); + snprintf (key, sizeof (key), "%s%d.restored_from_snap", prefix, count); + ret = dict_get_str (peer_data, key, &restored_snap); if (ret) { - gf_log (THIS->name, GF_LOG_ERROR, "Failed to get " - "is_volume_restored option for %s", - volname); + snprintf (msg, sizeof (msg), "%s missing in payload for %s", + key, volname); goto out; } - snprintf (key, sizeof (key), "volume%d.snap-max-hard-limit", count); - ret = dict_get_uint64 (vols, key, &new_volinfo->snap_max_hard_limit); + uuid_parse (restored_snap, new_volinfo->restored_from_snap); + + snprintf (key, sizeof (key), "%s%d.snap-max-hard-limit", prefix, count); + ret = dict_get_uint64 (peer_data, key, + &new_volinfo->snap_max_hard_limit); if (ret) { snprintf (msg, sizeof (msg), "%s missing in payload for %s", key, volname); @@ -3673,8 +3959,8 @@ glusterd_import_volinfo (dict_t *vols, int count, } memset (key, 0, sizeof (key)); - snprintf (key, sizeof (key), "volume%d.rebalance", count); - ret = dict_get_uint32 (vols, key, &new_volinfo->rebal.defrag_cmd); + snprintf (key, sizeof (key), "%s%d.rebalance", prefix, count); + ret = dict_get_uint32 (peer_data, key, &new_volinfo->rebal.defrag_cmd); if (ret) { snprintf (msg, sizeof (msg), "%s missing in payload for %s", key, volname); @@ -3682,8 +3968,8 @@ glusterd_import_volinfo (dict_t *vols, int count, } memset (key, 0, sizeof (key)); - snprintf (key, sizeof (key), "volume%d.rebalance-id", count); - ret = dict_get_str (vols, key, &rebalance_id_str); + snprintf (key, sizeof (key), "%s%d.rebalance-id", prefix, count); + ret = dict_get_str (peer_data, key, &rebalance_id_str); if (ret) { /* This is not present in older glusterfs versions, * so don't error out @@ -3694,15 +3980,17 @@ glusterd_import_volinfo (dict_t *vols, int count, } memset (key, 0, sizeof (key)); - snprintf (key, sizeof (key), "volume%d.rebalance-op", count); - ret = dict_get_uint32 (vols, key,(uint32_t *) &new_volinfo->rebal.op); + snprintf (key, sizeof (key), "%s%d.rebalance-op", prefix, count); + ret = dict_get_uint32 (peer_data, key, + (uint32_t *) &new_volinfo->rebal.op); if (ret) { /* This is not present in older glusterfs versions, * so don't error out */ ret = 0; } - ret = gd_import_friend_volume_rebal_dict (vols, count, new_volinfo); + ret = gd_import_friend_volume_rebal_dict (peer_data, count, + new_volinfo); if (ret) { snprintf (msg, sizeof (msg), "Failed to import rebalance dict " "for volume."); @@ -3710,8 +3998,8 @@ glusterd_import_volinfo (dict_t *vols, int count, } memset (key, 0, sizeof (key)); - snprintf (key, 256, "volume%d."GLUSTERD_STORE_KEY_RB_STATUS, count); - ret = dict_get_int32 (vols, key, &rb_status); + snprintf (key, 256, "%s%d."GLUSTERD_STORE_KEY_RB_STATUS, prefix, count); + ret = dict_get_int32 (peer_data, key, &rb_status); if (ret) goto out; new_volinfo->rep_brick.rb_status = rb_status; @@ -3719,9 +4007,9 @@ glusterd_import_volinfo (dict_t *vols, int count, if (new_volinfo->rep_brick.rb_status > GF_RB_STATUS_NONE) { memset (key, 0, sizeof (key)); - snprintf (key, 256, "volume%d."GLUSTERD_STORE_KEY_RB_SRC_BRICK, - count); - ret = dict_get_str (vols, key, &src_brick); + snprintf (key, 256, "%s%d."GLUSTERD_STORE_KEY_RB_SRC_BRICK, + prefix, count); + ret = dict_get_str (peer_data, key, &src_brick); if (ret) goto out; @@ -3734,9 +4022,9 @@ glusterd_import_volinfo (dict_t *vols, int count, } memset (key, 0, sizeof (key)); - snprintf (key, 256, "volume%d."GLUSTERD_STORE_KEY_RB_DST_BRICK, - count); - ret = dict_get_str (vols, key, &dst_brick); + snprintf (key, 256, "%s%d."GLUSTERD_STORE_KEY_RB_DST_BRICK, + prefix, count); + ret = dict_get_str (peer_data, key, &dst_brick); if (ret) goto out; @@ -3749,8 +4037,8 @@ glusterd_import_volinfo (dict_t *vols, int count, } memset (key, 0, sizeof (key)); - snprintf (key, sizeof (key), "volume%d.rb_id", count); - ret = dict_get_str (vols, key, &rb_id_str); + snprintf (key, sizeof (key), "%s%d.rb_id", prefix, count); + ret = dict_get_str (peer_data, key, &rb_id_str); if (ret) { /* This is not present in older glusterfs versions, * so don't error out @@ -3762,7 +4050,8 @@ glusterd_import_volinfo (dict_t *vols, int count, } - ret = glusterd_import_friend_volume_opts (vols, count, new_volinfo); + ret = glusterd_import_friend_volume_opts (peer_data, count, + new_volinfo); if (ret) goto out; @@ -3775,13 +4064,13 @@ glusterd_import_volinfo (dict_t *vols, int count, * present. Only one being present is a failure */ memset (key, 0, sizeof (key)); - snprintf (key, sizeof (key), "volume%d.op-version", count); - ret = dict_get_int32 (vols, key, &op_version); + snprintf (key, sizeof (key), "%s%d.op-version", prefix, count); + ret = dict_get_int32 (peer_data, key, &op_version); if (ret) ret = 0; memset (key, 0, sizeof (key)); - snprintf (key, sizeof (key), "volume%d.client-op-version", count); - ret = dict_get_int32 (vols, key, &client_op_version); + snprintf (key, sizeof (key), "%s%d.client-op-version", prefix, count); + ret = dict_get_int32 (peer_data, key, &client_op_version); if (ret) ret = 0; @@ -3800,11 +4089,11 @@ glusterd_import_volinfo (dict_t *vols, int count, } memset (key, 0 ,sizeof (key)); - snprintf (key, sizeof (key), "volume%d.caps", count); + snprintf (key, sizeof (key), "%s%d.caps", prefix, count); /*This is not present in older glusterfs versions, so ignore ret value*/ - ret = dict_get_int32 (vols, key, &new_volinfo->caps); + ret = dict_get_int32 (peer_data, key, &new_volinfo->caps); - ret = glusterd_import_bricks (vols, count, new_volinfo); + ret = glusterd_import_bricks (peer_data, count, new_volinfo, prefix); if (ret) goto out; @@ -3881,7 +4170,11 @@ glusterd_volinfo_stop_stale_bricks (glusterd_volinfo_t *new_volinfo, old_brickinfo->hostname, old_brickinfo->path, new_volinfo, &new_brickinfo); - if (ret) { + /* If the brick is stale, i.e it's not a part of the new volume + * or if it's part of the new volume and is pending a snap, + * then stop the brick process + */ + if (ret || (new_brickinfo->snap_status == -1)) { /*TODO: may need to switch to 'atomic' flavour of * brick_stop, once we make peer rpc program also * synctask enabled*/ @@ -3903,9 +4196,34 @@ int32_t glusterd_delete_stale_volume (glusterd_volinfo_t *stale_volinfo, glusterd_volinfo_t *valid_volinfo) { + int32_t ret = -1; + glusterd_volinfo_t *temp_volinfo = NULL; + glusterd_volinfo_t *voliter = NULL; + xlator_t *this = NULL; + GF_ASSERT (stale_volinfo); GF_ASSERT (valid_volinfo); + /* Copy snap_volumes list from stale_volinfo to valid_volinfo */ + valid_volinfo->snap_count = 0; + list_for_each_entry_safe (voliter, temp_volinfo, + &stale_volinfo->snap_volumes, snapvol_list) { + list_add_tail (&voliter->snapvol_list, + &valid_volinfo->snap_volumes); + valid_volinfo->snap_count++; + } + + if ((!uuid_is_null (stale_volinfo->restored_from_snap)) && + (uuid_compare (stale_volinfo->restored_from_snap, + valid_volinfo->restored_from_snap))) { + ret = glusterd_lvm_snapshot_remove (NULL, stale_volinfo); + if (ret) { + gf_log(this->name, GF_LOG_WARNING, + "Failed to remove lvm snapshot for " + "restored volume %s", stale_volinfo->volname); + } + } + /* If stale volume is in started state, copy the port numbers of the * local bricks if they exist in the valid volume information. * stop stale bricks. Stale volume information is going to be deleted. @@ -3992,7 +4310,7 @@ out: } int32_t -glusterd_import_friend_volume (dict_t *vols, size_t count) +glusterd_import_friend_volume (dict_t *peer_data, size_t count) { int32_t ret = -1; @@ -4001,13 +4319,14 @@ glusterd_import_friend_volume (dict_t *vols, size_t count) glusterd_volinfo_t *old_volinfo = NULL; glusterd_volinfo_t *new_volinfo = NULL; - GF_ASSERT (vols); + GF_ASSERT (peer_data); this = THIS; GF_ASSERT (this); priv = this->private; GF_ASSERT (priv); - ret = glusterd_import_volinfo (vols, count, &new_volinfo); + ret = glusterd_import_volinfo (peer_data, count, + &new_volinfo, "volume"); if (ret) goto out; @@ -4033,7 +4352,8 @@ glusterd_import_friend_volume (dict_t *vols, size_t count) if (ret) goto out; - ret = glusterd_import_quota_conf (vols, count, new_volinfo); + ret = glusterd_import_quota_conf (peer_data, count, + new_volinfo, "volume"); if (ret) goto out; @@ -4045,20 +4365,20 @@ out: } int32_t -glusterd_import_friend_volumes (dict_t *vols) +glusterd_import_friend_volumes (dict_t *peer_data) { int32_t ret = -1; int32_t count = 0; int i = 1; - GF_ASSERT (vols); + GF_ASSERT (peer_data); - ret = dict_get_int32 (vols, "count", &count); + ret = dict_get_int32 (peer_data, "count", &count); if (ret) goto out; while (i <= count) { - ret = glusterd_import_friend_volume (vols, i); + ret = glusterd_import_friend_volume (peer_data, i); if (ret) goto out; i++; @@ -4159,25 +4479,840 @@ out: } int32_t -glusterd_compare_friend_data (dict_t *vols, int32_t *status, char *hostname) +glusterd_perform_missed_op (glusterd_snap_t *snap, int32_t op) { - int32_t ret = -1; - int32_t count = 0; - int i = 1; - gf_boolean_t update = _gf_false; - gf_boolean_t stale_nfs = _gf_false; - gf_boolean_t stale_shd = _gf_false; - gf_boolean_t stale_qd = _gf_false; + dict_t *dict = NULL; + int32_t ret = -1; + glusterd_conf_t *priv = NULL; + glusterd_volinfo_t *snap_volinfo = NULL; + glusterd_volinfo_t *volinfo = NULL; + xlator_t *this = NULL; + uuid_t null_uuid = {0}; + + this = THIS; + GF_ASSERT (this); + + priv = this->private; + GF_ASSERT (priv); + GF_ASSERT (snap); + + dict = dict_new(); + if (!dict) { + gf_log (this->name, GF_LOG_ERROR, "Unable to create dict"); + ret = -1; + goto out; + } + + switch (op) { + case GF_SNAP_OPTION_TYPE_DELETE: + ret = glusterd_snap_remove (dict, snap, _gf_true, _gf_false); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, + "Failed to remove snap"); + goto out; + } + + break; + case GF_SNAP_OPTION_TYPE_RESTORE: + /* TODO : As of now there is only volume in snapshot. + * Change this when multiple volume snapshot is introduced + */ + snap_volinfo = list_entry (snap->volumes.next, + glusterd_volinfo_t, vol_list); + + /* Find the parent volinfo */ + ret = glusterd_volinfo_find (snap_volinfo->parent_volname, + &volinfo); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, + "Could not get volinfo of %s", + snap_volinfo->parent_volname); + goto out; + } + + /* Bump down the original volinfo's version, coz it would have + * incremented already due to volume handshake + */ + volinfo->version--; + uuid_copy (volinfo->restored_from_snap, null_uuid); + + /* Perform the restore */ + ret = gd_restore_snap_volume (dict, volinfo, snap_volinfo); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, "Failed to restore " + "snap for %s", snap->snapname); + volinfo->version++; + goto out; + } + + break; + default: + /* The entry must be a create, delete, or + * restore entry + */ + gf_log (this->name, GF_LOG_ERROR, "Invalid missed snap entry"); + ret = -1; + goto out; + } + +out: + dict_unref (dict); + gf_log (this->name, GF_LOG_TRACE, "Returning %d", ret); + return ret; +} + +/* Perform missed deletes and restores on this node */ +int32_t +glusterd_perform_missed_snap_ops () +{ + int32_t ret = -1; + int32_t op_status = -1; + glusterd_conf_t *priv = NULL; + glusterd_missed_snap_info *missed_snapinfo = NULL; + glusterd_snap_op_t *snap_opinfo = NULL; + glusterd_snap_t *snap = NULL; + uuid_t snap_uuid = {0,}; + xlator_t *this = NULL; + + this = THIS; + GF_ASSERT (this); + + priv = this->private; + GF_ASSERT (priv); + + list_for_each_entry (missed_snapinfo, &priv->missed_snaps_list, + missed_snaps) { + /* If the pending snap_op is not for this node then continue */ + if (strcmp (missed_snapinfo->node_uuid, uuid_utoa (MY_UUID))) + continue; + + /* Find the snap id */ + uuid_parse (missed_snapinfo->snap_uuid, snap_uuid); + snap = NULL; + snap = glusterd_find_snap_by_id (snap_uuid); + if (!snap) { + /* If the snap is not found, then a delete or a + * restore can't be pending on that snap_uuid. + */ + gf_log (this->name, GF_LOG_DEBUG, + "Not a pending delete or restore op"); + continue; + } - GF_ASSERT (vols); + op_status = GD_MISSED_SNAP_PENDING; + list_for_each_entry (snap_opinfo, &missed_snapinfo->snap_ops, + snap_ops_list) { + /* If the snap_op is create or its status is + * GD_MISSED_SNAP_DONE then continue + */ + if ((snap_opinfo->status == GD_MISSED_SNAP_DONE) || + (snap_opinfo->op == GF_SNAP_OPTION_TYPE_CREATE)) + continue; + + /* Perform the actual op for the first time for + * this snap, and mark the snap_status as + * GD_MISSED_SNAP_DONE. For other entries for the same + * snap, just mark the entry as done. + */ + if (op_status == GD_MISSED_SNAP_PENDING) { + ret = glusterd_perform_missed_op + (snap, + snap_opinfo->op); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, + "Failed to perform missed snap op"); + goto out; + } + op_status = GD_MISSED_SNAP_DONE; + } + + snap_opinfo->status = GD_MISSED_SNAP_DONE; + } + } + + ret = 0; +out: + gf_log (this->name, GF_LOG_TRACE, "Returning %d", ret); + return ret; +} + +/* Import friend volumes missed_snap_list and update * + * missed_snap_list if need be */ +int32_t +glusterd_import_friend_missed_snap_list (dict_t *peer_data) +{ + int32_t missed_snap_count = -1; + int32_t ret = -1; + glusterd_conf_t *priv = NULL; + xlator_t *this = NULL; + + this = THIS; + GF_ASSERT (this); + GF_ASSERT (peer_data); + + priv = this->private; + GF_ASSERT (priv); + + /* Add the friends missed_snaps entries to the in-memory list */ + ret = dict_get_int32 (peer_data, "missed_snap_count", + &missed_snap_count); + if (ret) { + gf_log (this->name, GF_LOG_INFO, + "No missed snaps"); + ret = 0; + goto out; + } + + ret = glusterd_add_missed_snaps_to_list (peer_data, + missed_snap_count); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, + "Failed to add missed snaps to list"); + goto out; + } + + ret = glusterd_perform_missed_snap_ops (); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, + "Failed to perform snap operations"); + /* Not going to out at this point coz some * + * missed ops might have been performed. We * + * need to persist the current list * + */ + } + + ret = glusterd_store_update_missed_snaps (); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, + "Failed to update missed_snaps_list"); + goto out; + } + +out: + gf_log (this->name, GF_LOG_TRACE, "Returning %d", ret); + return ret; +} + +/* Check for the peer_snap_name in the list of existing snapshots. + * If a snap exists with the same name and a different snap_id, then + * there is a conflict. Set conflict as _gf_true, and snap to the + * conflicting snap object. If a snap exists with the same name, and the + * same snap_id, then there is no conflict. Set conflict as _gf_false + * and snap to the existing snap object. If no snap exists with the + * peer_snap_name, then there is no conflict. Set conflict as _gf_false + * and snap to NULL. + */ +void +glusterd_is_peer_snap_conflicting (char *peer_snap_name, char *peer_snap_id, + gf_boolean_t *conflict, + glusterd_snap_t **snap, char *hostname) +{ + uuid_t peer_snap_uuid = {0,}; + xlator_t *this = NULL; + + this = THIS; + GF_ASSERT (this); + GF_ASSERT (peer_snap_name); + GF_ASSERT (peer_snap_id); + GF_ASSERT (conflict); + GF_ASSERT (snap); + GF_ASSERT (hostname); + + *snap = glusterd_find_snap_by_name (peer_snap_name); + if (*snap) { + uuid_parse (peer_snap_id, peer_snap_uuid); + if (!uuid_compare (peer_snap_uuid, (*snap)->snap_id)) { + /* Current node contains the same snap having + * the same snapname and snap_id + */ + gf_log (this->name, GF_LOG_DEBUG, + "Snapshot %s from peer %s present in " + "localhost", peer_snap_name, hostname); + *conflict = _gf_false; + } else { + /* Current node contains the same snap having + * the same snapname but different snap_id + */ + gf_log (this->name, GF_LOG_DEBUG, + "Snapshot %s from peer %s conflicts with " + "snapshot in localhost", peer_snap_name, + hostname); + *conflict = _gf_true; + } + } else { + /* Peer contains snapshots missing on the current node */ + gf_log (this->name, GF_LOG_INFO, + "Snapshot %s from peer %s missing on localhost", + peer_snap_name, hostname); + *conflict = _gf_false; + } +} + +/* Check if the local node is hosting any bricks for the given snapshot */ +gf_boolean_t +glusterd_are_snap_bricks_local (glusterd_snap_t *snap) +{ + gf_boolean_t is_local = _gf_false; + glusterd_volinfo_t *volinfo = NULL; + glusterd_brickinfo_t *brickinfo = NULL; + xlator_t *this = NULL; + + this = THIS; + GF_ASSERT (this); + GF_ASSERT (snap); + + list_for_each_entry (volinfo, &snap->volumes, vol_list) { + list_for_each_entry (brickinfo, &volinfo->bricks, brick_list) { + if (!uuid_compare (brickinfo->uuid, MY_UUID)) { + is_local = _gf_true; + goto out; + } + } + } + +out: + gf_log (this->name, GF_LOG_TRACE, "Returning %d", is_local); + return is_local; +} + +/* Check if the peer has missed any snap delete for the given snap_id */ +gf_boolean_t +glusterd_peer_has_missed_snap_delete (glusterd_peerinfo_t *peerinfo, + char *peer_snap_id) +{ + char *peer_uuid = NULL; + gf_boolean_t missed_delete = _gf_false; + glusterd_conf_t *priv = NULL; + glusterd_missed_snap_info *missed_snapinfo = NULL; + glusterd_snap_op_t *snap_opinfo = NULL; + xlator_t *this = NULL; + + this = THIS; + GF_ASSERT (this); + priv = this->private; + GF_ASSERT (priv); + GF_ASSERT (peerinfo); + GF_ASSERT (peer_snap_id); + + peer_uuid = uuid_utoa (peerinfo->uuid); + + list_for_each_entry (missed_snapinfo, &priv->missed_snaps_list, + missed_snaps) { + /* Look for missed snap for the same peer, and + * the same snap_id + */ + if ((!strcmp (peer_uuid, missed_snapinfo->node_uuid)) && + (!strcmp (peer_snap_id, missed_snapinfo->snap_uuid))) { + /* Check if the missed snap's op is delete and the + * status is pending + */ + list_for_each_entry (snap_opinfo, + &missed_snapinfo->snap_ops, + snap_ops_list) { + if ((snap_opinfo->op == + GF_SNAP_OPTION_TYPE_DELETE) && + (snap_opinfo->status == + GD_MISSED_SNAP_PENDING)) { + missed_delete = _gf_true; + goto out; + } + } + } + } + +out: + gf_log (this->name, GF_LOG_TRACE, "Returning %d", missed_delete); + return missed_delete; +} + +/* Genrate and store snap volfiles for imported snap object */ +int32_t +glusterd_gen_snap_volfiles (glusterd_volinfo_t *snap_vol, char *peer_snap_name) +{ + int32_t ret = -1; + xlator_t *this = NULL; + glusterd_volinfo_t *parent_volinfo = NULL; + glusterd_brickinfo_t *brickinfo = NULL; + + this = THIS; + GF_ASSERT (this); + GF_ASSERT (snap_vol); + GF_ASSERT (peer_snap_name); + + ret = glusterd_store_volinfo (snap_vol, GLUSTERD_VOLINFO_VER_AC_NONE); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, "Failed to store snapshot " + "volinfo (%s) for snap %s", snap_vol->volname, + peer_snap_name); + goto out; + } + + ret = generate_brick_volfiles (snap_vol); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, + "generating the brick volfiles for the " + "snap %s failed", peer_snap_name); + goto out; + } + + ret = generate_client_volfiles (snap_vol, GF_CLIENT_TRUSTED); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, + "generating the trusted client volfiles for " + "the snap %s failed", peer_snap_name); + goto out; + } + + ret = generate_client_volfiles (snap_vol, GF_CLIENT_OTHER); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, + "generating the client volfiles for the " + "snap %s failed", peer_snap_name); + goto out; + } + + ret = glusterd_volinfo_find (snap_vol->parent_volname, + &parent_volinfo); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, "Parent volinfo " + "not found for %s volume of snap %s", + snap_vol->volname, peer_snap_name); + goto out; + } + + glusterd_list_add_snapvol (parent_volinfo, snap_vol); + + list_for_each_entry (brickinfo, &snap_vol->bricks, brick_list) { + if (uuid_compare (brickinfo->uuid, MY_UUID)) + continue; + + if (brickinfo->snap_status == -1) { + gf_log (this->name, GF_LOG_INFO, + "not starting snap brick %s:%s for " + "for the snap %s (volume: %s)", + brickinfo->hostname, brickinfo->path, + peer_snap_name, parent_volinfo->volname); + continue; + } + + ret = glusterd_brick_start (snap_vol, brickinfo, _gf_true); + if (ret) { + gf_log (this->name, GF_LOG_WARNING, "starting the " + "brick %s:%s for the snap %s (volume: %s) " + "failed", brickinfo->hostname, brickinfo->path, + peer_snap_name, parent_volinfo->volname); + goto out; + } + } + + snap_vol->status = GLUSTERD_STATUS_STARTED; + + ret = glusterd_store_volinfo (snap_vol, GLUSTERD_VOLINFO_VER_AC_NONE); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, + "Failed to store snap volinfo"); + goto out; + } +out: + gf_log (this->name, GF_LOG_TRACE, "Returning %d", ret); + return ret; +} + +/* Import snapshot info from peer_data and add it to priv */ +int32_t +glusterd_import_friend_snap (dict_t *peer_data, int32_t snap_count, + char *peer_snap_name, char *peer_snap_id) +{ + char buf[NAME_MAX] = ""; + char prefix[NAME_MAX] = ""; + dict_t *dict = NULL; + glusterd_snap_t *snap = NULL; + glusterd_volinfo_t *snap_vol = NULL; + glusterd_conf_t *priv = NULL; + int32_t ret = -1; + int32_t volcount = -1; + int32_t i = -1; + xlator_t *this = NULL; + + this = THIS; + GF_ASSERT (this); + priv = this->private; + GF_ASSERT (priv); + GF_ASSERT (peer_data); + GF_ASSERT (peer_snap_name); + GF_ASSERT (peer_snap_id); + + snprintf (prefix, sizeof(prefix), "snap%d", snap_count); + + snap = glusterd_new_snap_object (); + if (!snap) { + gf_log (this->name, GF_LOG_ERROR, "Could not create " + "the snap object for snap %s", peer_snap_name); + goto out; + } + + strcpy (snap->snapname, peer_snap_name); + uuid_parse (peer_snap_id, snap->snap_id); + + snprintf (buf, sizeof(buf), "%s.snapid", prefix); + ret = dict_get_str (peer_data, buf, &snap->description); + + snprintf (buf, sizeof(buf), "%s.time_stamp", prefix); + ret = dict_get_int64 (peer_data, buf, &snap->time_stamp); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, + "Unable to get time_stamp for snap %s", + peer_snap_name); + goto out; + } + + snprintf (buf, sizeof(buf), "%s.snap_restored", prefix); + ret = dict_get_int8 (peer_data, buf, (int8_t *) &snap->snap_restored); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, + "Unable to get snap_restored for snap %s", + peer_snap_name); + goto out; + } + + snprintf (buf, sizeof(buf), "%s.snap_status", prefix); + ret = dict_get_int32 (peer_data, buf, (int32_t *) &snap->snap_status); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, + "Unable to get snap_status for snap %s", + peer_snap_name); + goto out; + } + + snprintf (buf, sizeof(buf), "%s.volcount", prefix); + ret = dict_get_int32 (peer_data, buf, &volcount); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, + "Unable to get volcount for snap %s", + peer_snap_name); + goto out; + } + + ret = glusterd_store_create_snap_dir (snap); + if (ret) { + gf_log (THIS->name, GF_LOG_ERROR, "Failed to create snap dir"); + goto out; + } + + list_add_order (&snap->snap_list, &priv->snapshots, + glusterd_compare_snap_time); + + for (i = 1; i <= volcount; i++) { + ret = glusterd_import_volinfo (peer_data, i, + &snap_vol, prefix); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, + "Failed to import snap volinfo for " + "snap %s", peer_snap_name); + goto out; + } + + snap_vol->snapshot = snap; + + ret = glusterd_gen_snap_volfiles (snap_vol, peer_snap_name); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, + "Failed to generate snap vol files " + "for snap %s", peer_snap_name); + goto out; + } + + ret = glusterd_import_quota_conf (peer_data, i, + snap_vol, prefix); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, + "Failed to import quota conf " + "for snap %s", peer_snap_name); + goto out; + } + + snap_vol = NULL; + } + + ret = glusterd_store_snap (snap); + if (ret) { + gf_log (this->name, GF_LOG_WARNING, "Could not store snap" + "object %s", peer_snap_name); + goto out; + } + +out: + if (ret) + glusterd_snap_remove (dict, snap, + _gf_true, _gf_true); + + if (dict) + dict_unref (dict); + + gf_log (this->name, GF_LOG_TRACE, "Returning %d", ret); + return ret; +} + +/* During a peer-handshake, after the volumes have synced, and the list of + * missed snapshots have synced, the node will perform the pending deletes + * and restores on this list. At this point, the current snapshot list in + * the node will be updated, and hence in case of conflicts arising during + * snapshot handshake, the peer hosting the bricks will be given precedence + * Likewise, if there will be a conflict, and both peers will be in the same + * state, i.e either both would be hosting bricks or both would not be hosting + * bricks, then a decision can't be taken and a peer-reject will happen. + * + * glusterd_compare_and_update_snap() implements the following algorithm to + * perform the above task: + * Step 1: Start. + * Step 2: Check if the peer is missing a delete on the said snap. + * If yes, goto step 6. + * Step 3: Check if there is a conflict between the peer's data and the + * local snap. If no, goto step 5. + * Step 4: As there is a conflict, check if both the peer and the local nodes + * are hosting bricks. Based on the results perform the following: + * Peer Hosts Bricks Local Node Hosts Bricks Action + * Yes Yes Goto Step 7 + * No No Goto Step 7 + * Yes No Goto Step 8 + * No Yes Goto Step 6 + * Step 5: Check if the local node is missing the peer's data. + * If yes, goto step 9. + * Step 6: It's a no-op. Goto step 10 + * Step 7: Peer Reject. Goto step 10 + * Step 8: Delete local node's data. + * Step 9: Accept Peer Data. + * Step 10: Stop + * + */ +int32_t +glusterd_compare_and_update_snap (dict_t *peer_data, int32_t snap_count, + glusterd_peerinfo_t *peerinfo) +{ + char buf[NAME_MAX] = ""; + char prefix[NAME_MAX] = ""; + char *peer_snap_name = NULL; + char *peer_snap_id = NULL; + dict_t *dict = NULL; + glusterd_snap_t *snap = NULL; + gf_boolean_t conflict = _gf_false; + gf_boolean_t is_local = _gf_false; + gf_boolean_t is_hosted = _gf_false; + gf_boolean_t missed_delete = _gf_false; + int32_t ret = -1; + xlator_t *this = NULL; + + this = THIS; + GF_ASSERT (this); + GF_ASSERT (peer_data); + GF_ASSERT (peerinfo); + + snprintf (prefix, sizeof(prefix), "snap%d", snap_count); + + /* Fetch the peer's snapname */ + snprintf (buf, sizeof(buf), "%s.snapname", prefix); + ret = dict_get_str (peer_data, buf, &peer_snap_name); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, + "Unable to fetch snapname from peer: %s", + peerinfo->hostname); + goto out; + } + + /* Fetch the peer's snap_id */ + snprintf (buf, sizeof(buf), "%s.snap_id", prefix); + ret = dict_get_str (peer_data, buf, &peer_snap_id); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, + "Unable to fetch snap_id from peer: %s", + peerinfo->hostname); + goto out; + } + + /* Check if the peer has missed a snap delete for the + * snap in question + */ + missed_delete = glusterd_peer_has_missed_snap_delete (peerinfo, + peer_snap_id); + if (missed_delete == _gf_true) { + /* Peer has missed delete on the missing/conflicting snap_id */ + gf_log (this->name, GF_LOG_INFO, "Peer %s has missed a delete " + "on snap %s", peerinfo->hostname, peer_snap_name); + ret = 0; + goto out; + } + + /* Check if there is a conflict, and if the + * peer data is already present + */ + glusterd_is_peer_snap_conflicting (peer_snap_name, peer_snap_id, + &conflict, &snap, + peerinfo->hostname); + if (conflict == _gf_false) { + if (snap) { + /* Peer has snap with the same snapname + * and snap_id. No need to accept peer data + */ + ret = 0; + goto out; + } else { + /* Peer has snap with the same snapname + * and snap_id, which local node doesn't have. + */ + goto accept_peer_data; + } + } + + /* There is a conflict. Check if the current node is + * hosting bricks for the conflicted snap. + */ + is_local = glusterd_are_snap_bricks_local (snap); + + /* Check if the peer is hosting any bricks for the + * conflicting snap + */ + snprintf (buf, sizeof(buf), "%s.host_bricks", prefix); + ret = dict_get_int8 (peer_data, buf, (int8_t *) &is_hosted); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, + "Unable to fetch host_bricks from peer: %s " + "for %s", peerinfo->hostname, peer_snap_name); + goto out; + } + + /* As there is a conflict at this point of time, the data of the + * node that hosts a brick takes precedence. If both the local + * node and the peer are in the same state, i.e if both of them + * are either hosting or not hosting the bricks, for the snap, + * then it's a peer reject + */ + if (is_hosted == is_local) { + gf_log (this->name, GF_LOG_ERROR, + "Conflict in snapshot %s with peer %s", + peer_snap_name, peerinfo->hostname); + ret = -1; + goto out; + } + + if (is_hosted == _gf_false) { + /* If there was a conflict, and the peer is not hosting + * any brick, then don't accept peer data + */ + gf_log (this->name, GF_LOG_DEBUG, + "Peer doesn't hosts bricks for conflicting " + "snap(%s). Not accepting peer data.", + peer_snap_name); + ret = 0; + goto out; + } + + /* The peer is hosting a brick in case of conflict + * And local node isn't. Hence remove local node's + * data and accept peer data + */ + + gf_log (this->name, GF_LOG_DEBUG, "Peer hosts bricks for conflicting " + "snap(%s). Removing local data. Accepting peer data.", + peer_snap_name); + + dict = dict_new(); + if (!dict) { + gf_log (this->name, GF_LOG_ERROR, + "Unable to create dict"); + ret = -1; + goto out; + } + + ret = glusterd_snap_remove (dict, snap, _gf_true, _gf_false); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, + "Failed to remove snap %s", snap->snapname); + goto out; + } + +accept_peer_data: + + /* Accept Peer Data */ + ret = glusterd_import_friend_snap (peer_data, snap_count, + peer_snap_name, peer_snap_id); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, + "Failed to import snap %s from peer %s", + peer_snap_name, peerinfo->hostname); + goto out; + } + +out: + if (dict) + dict_unref (dict); + + gf_log (this->name, GF_LOG_TRACE, "Returning %d", ret); + return ret; +} + +/* Compare snapshots present in peer_data, with the snapshots in + * the current node + */ +int32_t +glusterd_compare_friend_snapshots (dict_t *peer_data, + glusterd_peerinfo_t *peerinfo) +{ + int32_t ret = -1; + int32_t snap_count = 0; + int i = 1; + xlator_t *this = NULL; + + this = THIS; + GF_ASSERT (this); + GF_ASSERT (peer_data); + GF_ASSERT (peerinfo); + + ret = dict_get_int32 (peer_data, "snap_count", &snap_count); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, "Failed to fetch snap_count"); + goto out; + } + + for (i = 1; i <= snap_count; i++) { + /* Compare one snapshot from peer_data at a time */ + ret = glusterd_compare_and_update_snap (peer_data, i, peerinfo); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, + "Failed to compare snapshots with peer %s", + peerinfo->hostname); + goto out; + } + } + +out: + gf_log (this->name, GF_LOG_TRACE, "Returning %d", ret); + return ret; +} + +int32_t +glusterd_compare_friend_data (dict_t *peer_data, int32_t *status, + char *hostname) +{ + int32_t ret = -1; + int32_t count = 0; + int i = 1; + gf_boolean_t update = _gf_false; + gf_boolean_t stale_nfs = _gf_false; + gf_boolean_t stale_shd = _gf_false; + gf_boolean_t stale_qd = _gf_false; + xlator_t *this = NULL; + + this = THIS; + GF_ASSERT (this); + GF_ASSERT (peer_data); GF_ASSERT (status); - ret = dict_get_int32 (vols, "count", &count); + ret = dict_get_int32 (peer_data, "count", &count); if (ret) goto out; while (i <= count) { - ret = glusterd_compare_friend_volume (vols, i, status, + ret = glusterd_compare_friend_volume (peer_data, i, status, hostname); if (ret) goto out; @@ -4199,10 +5334,10 @@ glusterd_compare_friend_data (dict_t *vols, int32_t *status, char *hostname) stale_shd = _gf_true; if (glusterd_is_nodesvc_running ("quotad")) stale_qd = _gf_true; - ret = glusterd_import_global_opts (vols); + ret = glusterd_import_global_opts (peer_data); if (ret) goto out; - ret = glusterd_import_friend_volumes (vols); + ret = glusterd_import_friend_volumes (peer_data); if (ret) goto out; if (_gf_false == glusterd_are_all_volumes_stopped ()) { @@ -4218,9 +5353,8 @@ glusterd_compare_friend_data (dict_t *vols, int32_t *status, char *hostname) } out: - gf_log ("", GF_LOG_DEBUG, "Returning with ret: %d, status: %d", - ret, *status); - + gf_log (this->name, GF_LOG_DEBUG, + "Returning with ret: %d, status: %d", ret, *status); return ret; } @@ -4522,14 +5656,16 @@ glusterd_nodesvc_start (char *server, gf_boolean_t wait) "--trace-children=yes", "--track-origins=yes", NULL); runner_argprintf (&runner, "--log-file=%s", valgrind_logfile); - } + } runner_add_args (&runner, SBIN_DIR"/glusterfs", "-s", "localhost", "--volfile-id", volfileid, "-p", pidfile, "-l", logfile, - "-S", sockfpath, NULL); + "-S", sockfpath, + "-L", "DEBUG", + NULL); if (!strcmp (server, "glustershd")) { snprintf (glusterd_uuid_option, sizeof (glusterd_uuid_option), @@ -5456,7 +6592,6 @@ out: return -1; } -#ifdef GF_LINUX_HOST_OS int glusterd_get_brick_root (char *path, char **mount_point) { @@ -5752,7 +6887,6 @@ out: return device; } -#endif int glusterd_add_brick_detail_to_dict (glusterd_volinfo_t *volinfo, @@ -5826,13 +6960,12 @@ glusterd_add_brick_detail_to_dict (glusterd_volinfo_t *volinfo, if (ret) goto out; } -#ifdef GF_LINUX_HOST_OS + ret = glusterd_add_brick_mount_details (brickinfo, dict, count); if (ret) goto out; ret = glusterd_add_inode_size_to_dict (dict, count); -#endif out: if (ret) gf_log (this->name, GF_LOG_DEBUG, "Error adding brick" @@ -8830,7 +9963,7 @@ glusterd_snap_config_use_rsp_dict (dict_t *dst, dict_t *src) } for (i = 0; i < voldisplaycount; i++) { - snprintf (buf, sizeof(buf), "volume%ld-volname", i); + snprintf (buf, sizeof(buf), "volume%"PRIu64"-volname", i); ret = dict_get_str (src, buf, &volname); if (ret) { gf_log ("", GF_LOG_ERROR, @@ -8845,7 +9978,7 @@ glusterd_snap_config_use_rsp_dict (dict_t *dst, dict_t *src) } snprintf (buf, sizeof(buf), - "volume%ld-snap-max-hard-limit", i); + "volume%"PRIu64"-snap-max-hard-limit", i); ret = dict_get_uint64 (src, buf, &value); if (ret) { gf_log ("", GF_LOG_ERROR, @@ -8860,7 +9993,7 @@ glusterd_snap_config_use_rsp_dict (dict_t *dst, dict_t *src) } snprintf (buf, sizeof(buf), - "volume%ld-active-hard-limit", i); + "volume%"PRIu64"-active-hard-limit", i); ret = dict_get_uint64 (src, buf, &value); if (ret) { gf_log ("", GF_LOG_ERROR, @@ -8875,7 +10008,7 @@ glusterd_snap_config_use_rsp_dict (dict_t *dst, dict_t *src) } snprintf (buf, sizeof(buf), - "volume%ld-snap-max-soft-limit", i); + "volume%"PRIu64"-snap-max-soft-limit", i); ret = dict_get_uint64 (src, buf, &value); if (ret) { gf_log ("", GF_LOG_ERROR, @@ -9993,7 +11126,6 @@ glusterd_missed_snapinfo_new (glusterd_missed_snap_info **missed_snapinfo) if (!new_missed_snapinfo) goto out; - new_missed_snapinfo->node_snap_info = NULL; INIT_LIST_HEAD (&new_missed_snapinfo->missed_snaps); INIT_LIST_HEAD (&new_missed_snapinfo->snap_ops); @@ -10023,7 +11155,6 @@ glusterd_missed_snap_op_new (glusterd_snap_op_t **snap_op) if (!new_snap_op) goto out; - new_snap_op->brick_path = NULL; new_snap_op->brick_num = -1; new_snap_op->op = -1; new_snap_op->status = -1; @@ -10376,3 +11507,38 @@ glusterd_compare_volume_name(struct list_head *list1, struct list_head *list2) volinfo2 = list_entry(list2, glusterd_volinfo_t, vol_list); return strcmp(volinfo1->volname, volinfo2->volname); } + +int32_t +glusterd_mount_lvm_snapshot (char *device_path, char *brick_mount_path) +{ + char msg[NAME_MAX] = ""; + int32_t ret = -1; + runner_t runner = {0, }; + xlator_t *this = NULL; + + this = THIS; + GF_ASSERT (this); + GF_ASSERT (brick_mount_path); + GF_ASSERT (device_path); + + + runinit (&runner); + snprintf (msg, sizeof (msg), "mount -o nouuid %s %s", + device_path, brick_mount_path); + runner_add_args (&runner, "mount", "-o", "nouuid", device_path, + brick_mount_path, NULL); + runner_log (&runner, this->name, GF_LOG_DEBUG, msg); + ret = runner_run (&runner); + if (ret) { + gf_log (this->name, GF_LOG_ERROR, "mounting the snapshot " + "logical device %s failed (error: %s)", device_path, + strerror (errno)); + goto out; + } else + gf_log (this->name, GF_LOG_DEBUG, "mounting the snapshot " + "logical device %s successful", device_path); + +out: + gf_log (this->name, GF_LOG_TRACE, "Returning with %d", ret); + return ret; +} diff --git a/xlators/mgmt/glusterd/src/glusterd-utils.h b/xlators/mgmt/glusterd/src/glusterd-utils.h index 23f8ad7f6..84fa89b0e 100644 --- a/xlators/mgmt/glusterd/src/glusterd-utils.h +++ b/xlators/mgmt/glusterd/src/glusterd-utils.h @@ -8,7 +8,7 @@ cases as published by the Free Software Foundation. */ #ifndef _GLUSTERD_UTILS_H -#define _GLUSTERD_UTILS_H_ +#define _GLUSTERD_UTILS_H #ifndef _CONFIG_H #define _CONFIG_H @@ -182,10 +182,11 @@ glusterd_volume_brickinfo_get_by_brick (char *brick, glusterd_brickinfo_t **brickinfo); int32_t -glusterd_build_volume_dict (dict_t **vols); +glusterd_add_volumes_to_export_dict (dict_t **peer_data); int32_t -glusterd_compare_friend_data (dict_t *vols, int32_t *status, char *hostname); +glusterd_compare_friend_data (dict_t *peer_data, int32_t *status, + char *hostname); int glusterd_compute_cksum (glusterd_volinfo_t *volinfo, @@ -251,7 +252,7 @@ int glusterd_remote_hostname_get (rpcsvc_request_t *req, char *remote_host, int len); int32_t -glusterd_import_friend_volumes (dict_t *vols); +glusterd_import_friend_volumes (dict_t *peer_data); void glusterd_set_volume_status (glusterd_volinfo_t *volinfo, glusterd_volume_status status); @@ -280,7 +281,8 @@ int32_t glusterd_volume_count_get (void); int32_t glusterd_add_volume_to_dict (glusterd_volinfo_t *volinfo, - dict_t *dict, int32_t count); + dict_t *dict, int32_t count, + char *prefix); int glusterd_get_brickinfo (xlator_t *this, const char *brickname, int port, gf_boolean_t localhost, @@ -684,14 +686,16 @@ glusterd_rpc_clnt_unref (glusterd_conf_t *conf, rpc_clnt_t *rpc); int32_t glusterd_compare_volume_name(struct list_head *, struct list_head *); -#ifdef GF_LINUX_HOST_OS + char* glusterd_get_brick_mount_details (glusterd_brickinfo_t *brickinfo); + struct mntent * glusterd_get_mnt_entry_info (char *mnt_pt, FILE *mtab); + int glusterd_get_brick_root (char *path, char **mount_point); -#endif //LINUX_HOST + int glusterd_compare_snap_time(struct list_head *, struct list_head *); @@ -713,8 +717,42 @@ int32_t glusterd_missed_snap_op_new (glusterd_snap_op_t **snap_op); int32_t -glusterd_add_missed_snaps_to_dict (dict_t *rsp_dict, char *snap_uuid, +glusterd_add_missed_snaps_to_dict (dict_t *rsp_dict, + glusterd_volinfo_t *snap_vol, glusterd_brickinfo_t *brickinfo, int32_t brick_number, int32_t op); +int32_t +glusterd_add_missed_snaps_to_export_dict (dict_t *peer_data); + +int32_t +glusterd_import_friend_missed_snap_list (dict_t *peer_data); + +int32_t +gd_restore_snap_volume (dict_t *rsp_dict, + glusterd_volinfo_t *orig_vol, + glusterd_volinfo_t *snap_vol); + +int32_t +glusterd_mount_lvm_snapshot (char *device_path, char *brick_mount_path); + +int32_t +glusterd_add_snapshots_to_export_dict (dict_t *peer_data); + +int32_t +glusterd_compare_friend_snapshots (dict_t *peer_data, + glusterd_peerinfo_t *peerinfo); + +int32_t +glusterd_snapobject_delete (glusterd_snap_t *snap); + +int32_t +glusterd_snap_volume_remove (dict_t *rsp_dict, + glusterd_volinfo_t *snap_vol, + gf_boolean_t remove_lvm, + gf_boolean_t force); + +int32_t +glusterd_store_create_snap_dir (glusterd_snap_t *snap); + #endif diff --git a/xlators/mgmt/glusterd/src/glusterd-volgen.c b/xlators/mgmt/glusterd/src/glusterd-volgen.c index 332c3d359..f42d596ba 100644 --- a/xlators/mgmt/glusterd/src/glusterd-volgen.c +++ b/xlators/mgmt/glusterd/src/glusterd-volgen.c @@ -739,7 +739,7 @@ int glusterd_volinfo_get_boolean (glusterd_volinfo_t *volinfo, char *key) { char *val = NULL; - gf_boolean_t boo = _gf_false; + gf_boolean_t enabled = _gf_false; int ret = 0; ret = glusterd_volinfo_get (volinfo, key, &val); @@ -747,14 +747,14 @@ glusterd_volinfo_get_boolean (glusterd_volinfo_t *volinfo, char *key) return -1; if (val) - ret = gf_string2boolean (val, &boo); + ret = gf_string2boolean (val, &enabled); if (ret) { gf_log ("", GF_LOG_ERROR, "value for %s option is not valid", key); return -1; } - return boo; + return enabled; } gf_boolean_t @@ -1256,8 +1256,8 @@ static int server_check_marker_off (volgen_graph_t *graph, struct volopt_map_entry *vme, glusterd_volinfo_t *volinfo) { - gf_boolean_t bool = _gf_false; - int ret = 0; + gf_boolean_t enabled = _gf_false; + int ret = 0; GF_ASSERT (volinfo); GF_ASSERT (vme); @@ -1265,8 +1265,8 @@ server_check_marker_off (volgen_graph_t *graph, struct volopt_map_entry *vme, if (strcmp (vme->option, "!xtime") != 0) return 0; - ret = gf_string2boolean (vme->value, &bool); - if (ret || bool) + ret = gf_string2boolean (vme->value, &enabled); + if (ret || enabled) goto out; ret = glusterd_volinfo_get_boolean (volinfo, VKEY_MARKER_XTIME); @@ -1277,10 +1277,10 @@ server_check_marker_off (volgen_graph_t *graph, struct volopt_map_entry *vme, } if (ret) { - bool = _gf_false; - ret = glusterd_check_gsync_running (volinfo, &bool); + enabled = _gf_false; + ret = glusterd_check_gsync_running (volinfo, &enabled); - if (bool) { + if (enabled) { gf_log ("", GF_LOG_WARNING, GEOREP" sessions active" "for the volume %s, cannot disable marker " ,volinfo->volname); @@ -1900,6 +1900,10 @@ server_graph_builder (volgen_graph_t *graph, glusterd_volinfo_t *volinfo, if (ret) return -1; + xl = volgen_graph_add (graph, "features/barrier", volname); + if (!xl) + return -1; + ret = dict_get_int32 (volinfo->dict, "enable-pump", &pump); if (ret == -ENOENT) ret = pump = 0; @@ -4591,128 +4595,3 @@ gd_is_boolean_option (char *key) return _gf_false; } - -/* This function will restore origin volume to it's snap. - * The restore operation will simply replace the Gluster origin - * volume with the snap volume. - * TODO: Multi-volume delete to be done. - * Cleanup in case of restore failure is pending. - * - * @param orig_vol volinfo of origin volume - * @param snap_vol volinfo of snapshot volume - * - * @return 0 on success and negative value on error - */ -int -gd_restore_snap_volume (dict_t *rsp_dict, - glusterd_volinfo_t *orig_vol, - glusterd_volinfo_t *snap_vol) -{ - int ret = -1; - glusterd_volinfo_t *new_volinfo = NULL; - glusterd_snap_t *snap = NULL; - xlator_t *this = NULL; - glusterd_conf_t *conf = NULL; - glusterd_volinfo_t *temp_volinfo = NULL; - glusterd_volinfo_t *voliter = NULL; - - this = THIS; - GF_ASSERT (this); - GF_ASSERT (rsp_dict); - conf = this->private; - GF_ASSERT (conf); - - GF_VALIDATE_OR_GOTO (this->name, orig_vol, out); - GF_VALIDATE_OR_GOTO (this->name, snap_vol, out); - snap = snap_vol->snapshot; - GF_VALIDATE_OR_GOTO (this->name, snap, out); - - /* Snap volume must be stoped before performing the - * restore operation. - */ - ret = glusterd_stop_volume (snap_vol); - if (ret) { - gf_log (this->name, GF_LOG_ERROR, "Failed to stop " - "snap volume(%s)", snap_vol->volname); - goto out; - } - - /* Create a new volinfo for the restored volume */ - ret = glusterd_volinfo_dup (snap_vol, &new_volinfo, _gf_true); - if (ret) { - gf_log (this->name, GF_LOG_ERROR, "Failed to create volinfo"); - goto out; - } - - /* Following entries need to be derived from origin volume. */ - strcpy (new_volinfo->volname, orig_vol->volname); - uuid_copy (new_volinfo->volume_id, orig_vol->volume_id); - new_volinfo->snap_count = orig_vol->snap_count; - new_volinfo->snap_max_hard_limit = orig_vol->snap_max_hard_limit; - new_volinfo->is_volume_restored = _gf_true; - - /* Bump the version of the restored volume, so that nodes * - * which are done can sync during handshake */ - new_volinfo->version = orig_vol->version; - - list_for_each_entry_safe (voliter, temp_volinfo, - &orig_vol->snap_volumes, snapvol_list) { - list_add_tail (&voliter->snapvol_list, - &new_volinfo->snap_volumes); - } - /* Copy the snap vol info to the new_volinfo.*/ - ret = glusterd_snap_volinfo_restore (rsp_dict, new_volinfo, snap_vol); - if (ret) { - gf_log (this->name, GF_LOG_ERROR, "Failed to restore snap"); - (void)glusterd_volinfo_delete (new_volinfo); - goto out; - } - - /* If the orig_vol is already restored then we should delete - * the backend LVMs */ - if (orig_vol->is_volume_restored) { - ret = glusterd_lvm_snapshot_remove (rsp_dict, orig_vol); - if (ret) { - gf_log (this->name, GF_LOG_ERROR, "Failed to remove " - "LVM backend"); - (void)glusterd_volinfo_delete (new_volinfo); - goto out; - } - } - - /* Once the new_volinfo is completely constructed then delete - * the orinal volinfo - */ - ret = glusterd_volinfo_delete (orig_vol); - if (ret) { - gf_log (this->name, GF_LOG_ERROR, "Failed to delete volinfo"); - (void)glusterd_volinfo_delete (new_volinfo); - goto out; - } - /* New volinfo always shows the status as created. Therefore - * set the status to stop. */ - glusterd_set_volume_status (new_volinfo, GLUSTERD_STATUS_STOPPED); - - list_add_tail (&new_volinfo->vol_list, &conf->volumes); - - /* Now delete the snap entry. As a first step delete the snap - * volume information stored in store. */ - ret = glusterd_snap_remove (rsp_dict, snap, _gf_false, _gf_true); - if (ret) { - gf_log (this->name, GF_LOG_WARNING, "Failed to delete " - "snap %s", snap->snapname); - goto out; - } - - ret = glusterd_store_volinfo (new_volinfo, - GLUSTERD_VOLINFO_VER_AC_INCREMENT); - if (ret) { - gf_log (this->name, GF_LOG_ERROR, "Failed to store volinfo"); - goto out; - } - - ret = 0; -out: - - return ret; -} diff --git a/xlators/mgmt/glusterd/src/glusterd-volgen.h b/xlators/mgmt/glusterd/src/glusterd-volgen.h index dcba11f32..f4703c288 100644 --- a/xlators/mgmt/glusterd/src/glusterd-volgen.h +++ b/xlators/mgmt/glusterd/src/glusterd-volgen.h @@ -176,7 +176,4 @@ gd_is_xlator_option (char *key); gf_boolean_t gd_is_boolean_option (char *key); -int gd_restore_snap_volume (dict_t *rsp_dict, - glusterd_volinfo_t *orig_vol, - glusterd_volinfo_t *snap_vol); #endif diff --git a/xlators/mgmt/glusterd/src/glusterd-volume-ops.c b/xlators/mgmt/glusterd/src/glusterd-volume-ops.c index eac926d95..504aeb839 100644 --- a/xlators/mgmt/glusterd/src/glusterd-volume-ops.c +++ b/xlators/mgmt/glusterd/src/glusterd-volume-ops.c @@ -1213,7 +1213,7 @@ glusterd_op_stage_delete_volume (dict_t *dict, char **op_errstr) if (volinfo->snap_count > 0 || !list_empty(&volinfo->snap_volumes)) { snprintf (msg, sizeof (msg), "Cannot delete Volume %s ," - "as it has %ld snapshots. " + "as it has %"PRIu64" snapshots. " "To delete the volume, " "first delete all the snapshots under it.", volname, volinfo->snap_count); diff --git a/xlators/mgmt/glusterd/src/glusterd-volume-set.c b/xlators/mgmt/glusterd/src/glusterd-volume-set.c index 3f1d7d539..1374e82cd 100644 --- a/xlators/mgmt/glusterd/src/glusterd-volume-set.c +++ b/xlators/mgmt/glusterd/src/glusterd-volume-set.c @@ -116,8 +116,8 @@ validate_cache_max_min_size (dict_t *dict, char *key, char *value, "performance.cache-max-file-size", ¤t_max_value); if (current_max_value) { - gf_string2bytesize (current_max_value, &max_value); - gf_string2bytesize (value, &min_value); + gf_string2bytesize_uint64 (current_max_value, &max_value); + gf_string2bytesize_uint64 (value, &min_value); current_min_value = value; } } else if ((!strcmp (key, "performance.cache-max-file-size")) || @@ -126,8 +126,8 @@ validate_cache_max_min_size (dict_t *dict, char *key, char *value, "performance.cache-min-file-size", ¤t_min_value); if (current_min_value) { - gf_string2bytesize (current_min_value, &min_value); - gf_string2bytesize (value, &max_value); + gf_string2bytesize_uint64 (current_min_value, &min_value); + gf_string2bytesize_uint64 (value, &max_value); current_max_value = value; } } @@ -1389,6 +1389,18 @@ struct volopt_map_entry glusterd_volopt_map[] = { .type = GLOBAL_DOC, .op_version = 1 }, + { .key = "nfs.rpc-statd", + .voltype = "nfs/server", + .option = "nfs.rpc-statd", + .type = NO_DOC, + .op_version = 4, + }, + { .key = "nfs.log-level", + .voltype = "nfs/server", + .option = "nfs.log-level", + .type = NO_DOC, + .op_version = 4, + }, { .key = "nfs.server-aux-gids", .voltype = "nfs/server", .option = "nfs.server-aux-gids", @@ -1453,6 +1465,10 @@ struct volopt_map_entry glusterd_volopt_map[] = { .voltype = "storage/posix", .op_version = 3 }, + { .key = "storage.xattr-user-namespace-mode", + .voltype = "storage/posix", + .op_version = 4 + }, { .key = "storage.owner-uid", .voltype = "storage/posix", .option = "brick-uid", @@ -1527,6 +1543,16 @@ struct volopt_map_entry glusterd_volopt_map[] = { .type = NO_DOC, .op_version = 3 }, + { .key = "features.barrier", + .voltype = "features/barrier", + .value = "disable", + .op_version = 4 + }, + { .key = "features.barrier-timeout", + .voltype = "features/barrier", + .value = "120", + .op_version = 4 + }, { .key = NULL } }; diff --git a/xlators/mgmt/glusterd/src/glusterd.c b/xlators/mgmt/glusterd/src/glusterd.c index 6e7a9278d..4d09d7fd9 100644 --- a/xlators/mgmt/glusterd/src/glusterd.c +++ b/xlators/mgmt/glusterd/src/glusterd.c @@ -30,6 +30,7 @@ #include "dict.h" #include "compat.h" #include "compat-errno.h" +#include "syscall.h" #include "statedump.h" #include "glusterd-sm.h" #include "glusterd-op-sm.h" @@ -311,7 +312,7 @@ out: } -inline int32_t +static inline int32_t glusterd_program_register (xlator_t *this, rpcsvc_t *svc, rpcsvc_program_t *prog) { @@ -804,7 +805,7 @@ check_prepare_mountbroker_root (char *mountbroker_root) dfd0 = dup (dfd); for (;;) { - ret = openat (dfd, "..", O_RDONLY); + ret = sys_openat (dfd, "..", O_RDONLY); if (ret != -1) { dfd2 = ret; ret = fstat (dfd2, &st2); @@ -839,11 +840,11 @@ check_prepare_mountbroker_root (char *mountbroker_root) st = st2; } - ret = mkdirat (dfd0, MB_HIVE, 0711); + ret = sys_mkdirat (dfd0, MB_HIVE, 0711); if (ret == -1 && errno == EEXIST) ret = 0; if (ret != -1) - ret = fstatat (dfd0, MB_HIVE, &st, AT_SYMLINK_NOFOLLOW); + ret = sys_fstatat (dfd0, MB_HIVE, &st, AT_SYMLINK_NOFOLLOW); if (ret == -1 || st.st_mode != (S_IFDIR|0711)) { gf_log ("", GF_LOG_ERROR, "failed to set up mountbroker-root directory %s", @@ -1255,6 +1256,17 @@ init (xlator_t *this) exit (1); } + snprintf (storedir, PATH_MAX, "%s/snaps", workdir); + + ret = mkdir (storedir, 0777); + + if ((-1 == ret) && (errno != EEXIST)) { + gf_log (this->name, GF_LOG_CRITICAL, + "Unable to create snaps directory %s" + " ,errno = %d", storedir, errno); + exit (1); + } + snprintf (storedir, PATH_MAX, "%s/peers", workdir); ret = mkdir (storedir, 0777); diff --git a/xlators/mgmt/glusterd/src/glusterd.h b/xlators/mgmt/glusterd/src/glusterd.h index 79461fb5d..7157bee64 100644 --- a/xlators/mgmt/glusterd/src/glusterd.h +++ b/xlators/mgmt/glusterd/src/glusterd.h @@ -310,7 +310,7 @@ struct glusterd_volinfo_ { char volname[GLUSTERD_MAX_VOLUME_NAME]; gf_boolean_t is_snap_volume; glusterd_snap_t *snapshot; - gf_boolean_t is_volume_restored; + uuid_t restored_from_snap; char parent_volname[GLUSTERD_MAX_VOLUME_NAME]; /* In case of a snap volume i.e (is_snap_volume == TRUE) this @@ -405,6 +405,7 @@ struct glusterd_snap_ { }; typedef struct glusterd_snap_op_ { + char *snap_vol_id; int32_t brick_num; char *brick_path; int32_t op; @@ -413,7 +414,8 @@ typedef struct glusterd_snap_op_ { } glusterd_snap_op_t; typedef struct glusterd_missed_snap_ { - char *node_snap_info; + char *node_uuid; + char *snap_uuid; struct list_head missed_snaps; struct list_head snap_ops; } glusterd_missed_snap_info; @@ -1010,8 +1012,8 @@ int32_t glusterd_add_missed_snaps_to_list (dict_t *dict, int32_t missed_snap_count); int32_t -glusterd_store_missed_snaps_list (char *missed_info, int32_t brick_num, - char *brick_path, int32_t snap_op, - int32_t snap_status); +glusterd_add_new_entry_to_list (char *missed_info, char *snap_vol_id, + int32_t brick_num, char *brick_path, + int32_t snap_op, int32_t snap_status); #endif |