From 5a4f4a945661a8bb24735524e152ccd5b1ba571a Mon Sep 17 00:00:00 2001 From: Atin Mukherjee Date: Wed, 11 May 2016 18:24:40 +0530 Subject: glusterd: copy real_path from older brickinfo during brick import In glusterd_import_new_brick () new_brickinfo->real_path will not be populated for the first time and hence if the underlying file system is bad for the same brick, import will fail resulting in inconsistent configuration data. Fix is to populate real_path from old brickinfo object. Also there were many cases where we were unnecessarily calling realpath() and that may cause in failure. For eg - if a remove brick is executed with a brick whoose underlying file system has crashed, remove-brick fails since realpath() call fails. We'd need to call realpath() here as the value is of no use.Hence passing construct_realpath as _gf_false in glusterd_volume_brickinfo_get_by_brick () is a must in such cases. Change-Id: I7ec93871dc9e616f5d565ad5e540b2f1cacaf9dc BUG: 1335531 Signed-off-by: Atin Mukherjee Reviewed-on: http://review.gluster.org/14306 Smoke: Gluster Build System NetBSD-regression: NetBSD Build System CentOS-regression: Gluster Build System Reviewed-by: Kaushal M --- xlators/mgmt/glusterd/src/glusterd-brick-ops.c | 10 +-- xlators/mgmt/glusterd/src/glusterd-log-ops.c | 4 +- xlators/mgmt/glusterd/src/glusterd-op-sm.c | 8 +- xlators/mgmt/glusterd/src/glusterd-rebalance.c | 2 +- xlators/mgmt/glusterd/src/glusterd-replace-brick.c | 10 +-- xlators/mgmt/glusterd/src/glusterd-store.c | 3 +- xlators/mgmt/glusterd/src/glusterd-utils.c | 95 ++++++++++++++-------- xlators/mgmt/glusterd/src/glusterd-utils.h | 3 +- xlators/mgmt/glusterd/src/glusterd-volgen.c | 3 +- xlators/mgmt/glusterd/src/glusterd-volume-ops.c | 4 +- 10 files changed, 86 insertions(+), 56 deletions(-) (limited to 'xlators') diff --git a/xlators/mgmt/glusterd/src/glusterd-brick-ops.c b/xlators/mgmt/glusterd/src/glusterd-brick-ops.c index 35c834a11ce..5b2f559b546 100644 --- a/xlators/mgmt/glusterd/src/glusterd-brick-ops.c +++ b/xlators/mgmt/glusterd/src/glusterd-brick-ops.c @@ -1035,7 +1035,7 @@ __glusterd_handle_remove_brick (rpcsvc_request_t *req) ret = glusterd_volume_brickinfo_get_by_brick(brick, volinfo, &brickinfo, - _gf_true); + _gf_false); if (ret) { snprintf (err_str, sizeof (err_str), "Incorrect brick " @@ -1276,7 +1276,7 @@ glusterd_op_perform_add_bricks (glusterd_volinfo_t *volinfo, int32_t count, goto out; while ( i <= count) { ret = glusterd_brickinfo_new_from_brick (brick, &brickinfo, - _gf_true); + _gf_true, NULL); if (ret) goto out; @@ -1488,7 +1488,7 @@ glusterd_op_perform_remove_brick (glusterd_volinfo_t *volinfo, char *brick, ret = glusterd_volume_brickinfo_get_by_brick (brick, volinfo, &brickinfo, - _gf_true); + _gf_false); if (ret) goto out; @@ -1704,7 +1704,7 @@ glusterd_op_stage_add_brick (dict_t *dict, char **op_errstr, dict_t *rsp_dict) } ret = glusterd_brickinfo_new_from_brick (brick, &brickinfo, - _gf_true); + _gf_true, NULL); if (ret) { gf_msg (THIS->name, GF_LOG_ERROR, 0, GD_MSG_BRICK_NOT_FOUND, @@ -1828,7 +1828,7 @@ glusterd_remove_brick_validate_bricks (gf1_op_commands cmd, int32_t brick_count, ret = glusterd_volume_brickinfo_get_by_brick(brick, volinfo, &brickinfo, - _gf_true); + _gf_false); if (ret) { snprintf (msg, sizeof (msg), "Incorrect brick " "%s for volume %s", brick, volinfo->volname); diff --git a/xlators/mgmt/glusterd/src/glusterd-log-ops.c b/xlators/mgmt/glusterd/src/glusterd-log-ops.c index 4d66df46132..d04492af7cc 100644 --- a/xlators/mgmt/glusterd/src/glusterd-log-ops.c +++ b/xlators/mgmt/glusterd/src/glusterd-log-ops.c @@ -147,7 +147,7 @@ glusterd_op_stage_log_rotate (dict_t *dict, char **op_errstr) } ret = glusterd_volume_brickinfo_get_by_brick (brick, volinfo, NULL, - _gf_true); + _gf_false); if (ret) { snprintf (msg, sizeof (msg), "Incorrect brick %s " "for volume %s", brick, volname); @@ -207,7 +207,7 @@ glusterd_op_log_rotate (dict_t *dict) goto cont; ret = glusterd_brickinfo_new_from_brick (brick, &tmpbrkinfo, - _gf_true); + _gf_false, NULL); if (ret) { gf_msg ("glusterd", GF_LOG_ERROR, 0, GD_MSG_BRICK_NOT_FOUND, diff --git a/xlators/mgmt/glusterd/src/glusterd-op-sm.c b/xlators/mgmt/glusterd/src/glusterd-op-sm.c index 033fc5759f2..e9f261c2fb3 100644 --- a/xlators/mgmt/glusterd/src/glusterd-op-sm.c +++ b/xlators/mgmt/glusterd/src/glusterd-op-sm.c @@ -1724,7 +1724,7 @@ glusterd_op_stage_status_volume (dict_t *dict, char **op_errstr) ret = glusterd_volume_brickinfo_get_by_brick (brick, volinfo, &brickinfo, - _gf_true); + _gf_false); if (ret) { snprintf (msg, sizeof(msg), "No brick %s in" " volume %s", brick, volname); @@ -3242,7 +3242,7 @@ glusterd_op_status_volume (dict_t *dict, char **op_errstr, ret = glusterd_volume_brickinfo_get_by_brick (brick, volinfo, &brickinfo, - _gf_true); + _gf_false); if (ret) goto out; @@ -5953,7 +5953,7 @@ glusterd_bricks_select_remove_brick (dict_t *dict, char **op_errstr, ret = glusterd_volume_brickinfo_get_by_brick (brick, volinfo, &brickinfo, - _gf_true); + _gf_false); if (ret) goto out; @@ -6813,7 +6813,7 @@ glusterd_bricks_select_status_volume (dict_t *dict, char **op_errstr, ret = glusterd_volume_brickinfo_get_by_brick (brickname, volinfo, &brickinfo, - _gf_true); + _gf_false); if (ret) goto out; diff --git a/xlators/mgmt/glusterd/src/glusterd-rebalance.c b/xlators/mgmt/glusterd/src/glusterd-rebalance.c index 60d1b204da5..fff6b3c2d91 100644 --- a/xlators/mgmt/glusterd/src/glusterd-rebalance.c +++ b/xlators/mgmt/glusterd/src/glusterd-rebalance.c @@ -603,7 +603,7 @@ glusterd_brick_validation (dict_t *dict, char *key, data_t *value, ret = glusterd_volume_brickinfo_get_by_brick (value->data, volinfo, &brickinfo, - _gf_true); + _gf_false); if (ret) { gf_msg (this->name, GF_LOG_ERROR, EINVAL, GD_MSG_BRICK_NOT_FOUND, diff --git a/xlators/mgmt/glusterd/src/glusterd-replace-brick.c b/xlators/mgmt/glusterd/src/glusterd-replace-brick.c index 30e6ee850e9..2b2120738ca 100644 --- a/xlators/mgmt/glusterd/src/glusterd-replace-brick.c +++ b/xlators/mgmt/glusterd/src/glusterd-replace-brick.c @@ -295,7 +295,7 @@ glusterd_op_stage_replace_brick (dict_t *dict, char **op_errstr, ret = glusterd_volume_brickinfo_get_by_brick (src_brick, volinfo, &src_brickinfo, - _gf_true); + _gf_false); if (ret) { snprintf (msg, sizeof (msg), "brick: %s does not exist in " "volume: %s", src_brick, volname); @@ -365,7 +365,7 @@ glusterd_op_stage_replace_brick (dict_t *dict, char **op_errstr, } ret = glusterd_brickinfo_new_from_brick (dst_brick, &dst_brickinfo, - _gf_true); + _gf_true, NULL); if (ret) goto out; @@ -551,7 +551,7 @@ glusterd_op_perform_replace_brick (glusterd_volinfo_t *volinfo, GF_ASSERT (conf); ret = glusterd_brickinfo_new_from_brick (new_brick, &new_brickinfo, - _gf_true); + _gf_true, NULL); if (ret) goto out; @@ -562,7 +562,7 @@ glusterd_op_perform_replace_brick (glusterd_volinfo_t *volinfo, ret = glusterd_volume_brickinfo_get_by_brick (old_brick, volinfo, &old_brickinfo, - _gf_true); + _gf_false); if (ret) goto out; @@ -682,7 +682,7 @@ glusterd_op_replace_brick (dict_t *dict, dict_t *rsp_dict) ret = glusterd_volume_brickinfo_get_by_brick (src_brick, volinfo, &src_brickinfo, - _gf_true); + _gf_false); if (ret) { gf_msg_debug (this->name, 0, "Unable to get src-brickinfo"); diff --git a/xlators/mgmt/glusterd/src/glusterd-store.c b/xlators/mgmt/glusterd/src/glusterd-store.c index aa1201e492d..113902d2e6a 100644 --- a/xlators/mgmt/glusterd/src/glusterd-store.c +++ b/xlators/mgmt/glusterd/src/glusterd-store.c @@ -165,7 +165,8 @@ glusterd_store_is_valid_brickpath (char *volname, char *brick) this = THIS; GF_ASSERT (this); - ret = glusterd_brickinfo_new_from_brick (brick, &brickinfo, _gf_true); + ret = glusterd_brickinfo_new_from_brick (brick, &brickinfo, _gf_false, + NULL); if (ret) { gf_msg (this->name, GF_LOG_WARNING, 0, GD_MSG_BRICK_CREATION_FAIL, "Failed to create brick " diff --git a/xlators/mgmt/glusterd/src/glusterd-utils.c b/xlators/mgmt/glusterd/src/glusterd-utils.c index 41e90c2faeb..89c6d2dfa18 100644 --- a/xlators/mgmt/glusterd/src/glusterd-utils.c +++ b/xlators/mgmt/glusterd/src/glusterd-utils.c @@ -1061,7 +1061,8 @@ out: int32_t glusterd_brickinfo_new_from_brick (char *brick, glusterd_brickinfo_t **brickinfo, - gf_boolean_t construct_real_path) + gf_boolean_t construct_real_path, + char **op_errstr) { char *hostname = NULL; char *path = NULL; @@ -1109,7 +1110,25 @@ glusterd_brickinfo_new_from_brick (char *brick, strncpy (new_brickinfo->hostname, hostname, 1024); strncpy (new_brickinfo->path, path, 1024); - if (construct_real_path && new_brickinfo->real_path[0] == '\0') { + if (construct_real_path) { + ret = glusterd_hostname_to_uuid (new_brickinfo->hostname, + new_brickinfo->uuid); + if (ret) { + gf_msg (this->name, GF_LOG_ERROR, 0, + GD_MSG_HOSTNAME_TO_UUID_FAIL, + "Failed to convert hostname %s to uuid", + hostname); + if (op_errstr) + gf_asprintf (op_errstr, "Host %s is not in \' " + "Peer in Cluster\' state", + new_brickinfo->hostname); + goto out; + } + } + + if (construct_real_path && + !gf_uuid_compare (new_brickinfo->uuid, MY_UUID) + && new_brickinfo->real_path[0] == '\0') { if (!realpath (new_brickinfo->path, abspath)) { /* ENOENT indicates that brick path has not been created * which is a valid scenario */ @@ -1435,7 +1454,7 @@ glusterd_volume_brickinfo_get_by_brick (char *brick, GF_ASSERT (volinfo); ret = glusterd_brickinfo_new_from_brick (brick, &tmp_brickinfo, - construct_real_path); + construct_real_path, NULL); if (ret) goto out; @@ -3122,12 +3141,8 @@ glusterd_import_new_brick (dict_t *peer_data, int32_t vol_count, int decommissioned = 0; glusterd_brickinfo_t *new_brickinfo = NULL; char msg[2048] = {0}; - xlator_t *this = NULL; char *brick_uuid_str = NULL; - char abspath[PATH_MAX] = {0}; - this = THIS; - GF_ASSERT (this); GF_ASSERT (peer_data); GF_ASSERT (vol_count >= 0); GF_ASSERT (brickinfo); @@ -3188,23 +3203,7 @@ glusterd_import_new_brick (dict_t *peer_data, int32_t vol_count, ret = dict_get_str (peer_data, key, &brick_uuid_str); if (ret) goto out; - gf_uuid_parse (brick_uuid_str, new_brickinfo->uuid); - if (!gf_uuid_compare(new_brickinfo->uuid, MY_UUID)) { - if (new_brickinfo->real_path[0] == '\0') { - if (!realpath (new_brickinfo->path, abspath)) { - gf_msg (this->name, GF_LOG_CRITICAL, errno, - GD_MSG_BRICKINFO_CREATE_FAIL, - "realpath() failed for brick %s. The " - "underlying file system may be in bad " - "state", new_brickinfo->path); - ret = -1; - goto out; - } - strncpy (new_brickinfo->real_path, abspath, - strlen(abspath)); - } - } *brickinfo = new_brickinfo; out: @@ -3219,7 +3218,7 @@ out: * It will be "volume" for normal volumes, and snap# like * snap1, snap2, for snapshot volumes */ -int32_t +static int32_t glusterd_import_bricks (dict_t *peer_data, int32_t vol_count, glusterd_volinfo_t *new_volinfo, char *prefix) { @@ -3429,6 +3428,7 @@ glusterd_import_volinfo (dict_t *peer_data, int count, char *parent_volname = NULL; char *volname = NULL; glusterd_volinfo_t *new_volinfo = NULL; + glusterd_volinfo_t *old_volinfo = NULL; char *volume_id_str = NULL; char *restored_snap = NULL; char msg[2048] = {0}; @@ -3847,8 +3847,8 @@ glusterd_volume_disconnect_all_bricks (glusterd_volinfo_t *volinfo) } int32_t -glusterd_volinfo_copy_brick_portinfo (glusterd_volinfo_t *old_volinfo, - glusterd_volinfo_t *new_volinfo) +glusterd_volinfo_copy_brickinfo (glusterd_volinfo_t *old_volinfo, + glusterd_volinfo_t *new_volinfo) { char pidfile[PATH_MAX+1] = {0,}; glusterd_brickinfo_t *new_brickinfo = NULL; @@ -3856,6 +3856,7 @@ glusterd_volinfo_copy_brick_portinfo (glusterd_volinfo_t *old_volinfo, glusterd_conf_t *priv = NULL; int ret = 0; xlator_t *this = NULL; + char abspath[PATH_MAX] = {0}; GF_ASSERT (new_volinfo); GF_ASSERT (old_volinfo); @@ -3873,10 +3874,35 @@ glusterd_volinfo_copy_brick_portinfo (glusterd_volinfo_t *old_volinfo, &old_brickinfo); if (ret == 0) { new_brickinfo->port = old_brickinfo->port; + + if (old_brickinfo->real_path == '\0') { + if (!realpath (new_brickinfo->path, abspath)) { + /* Here an ENOENT should also be a + * failure as the brick is expected to + * be in existance + */ + gf_msg (this->name, GF_LOG_CRITICAL, + errno, + GD_MSG_BRICKINFO_CREATE_FAIL, + "realpath () failed for brick " + "%s. The underlying filesystem " + "may be in bad state", + new_brickinfo->path); + ret = -1; + goto out; + } + strncpy (new_brickinfo->real_path, abspath, + strlen(abspath)); + } else { + strncpy (new_brickinfo->real_path, + old_brickinfo->real_path, + strlen (old_brickinfo->real_path)); + } } } ret = 0; +out: return ret; } @@ -4060,8 +4086,8 @@ glusterd_import_friend_volume (dict_t *peer_data, size_t count) glusterd_volinfo_t *old_volinfo = NULL; glusterd_volinfo_t *new_volinfo = NULL; glusterd_svc_t *svc = NULL; - gf_boolean_t newexportvalue; - gf_boolean_t oldexportvalue; + gf_boolean_t newexportvalue = _gf_false; + gf_boolean_t oldexportvalue = _gf_false; char *value = NULL; GF_ASSERT (peer_data); @@ -4092,11 +4118,12 @@ glusterd_import_friend_volume (dict_t *peer_data, size_t count) (void) gd_check_and_update_rebalance_info (old_volinfo, new_volinfo); - /* Copy brick ports from the old volinfo always. The old_volinfo - * will be cleaned up and this information could be lost + /* Copy brick ports & real_path from the old volinfo always. + * The old_volinfo will be cleaned up and this information + * could be lost */ - (void) glusterd_volinfo_copy_brick_portinfo (old_volinfo, - new_volinfo); + (void) glusterd_volinfo_copy_brickinfo (old_volinfo, + new_volinfo); (void) glusterd_delete_stale_volume (old_volinfo, new_volinfo); glusterd_volinfo_unref (old_volinfo); @@ -5913,7 +5940,7 @@ glusterd_new_brick_validate (char *brick, glusterd_brickinfo_t *brickinfo, if (!brickinfo) { ret = glusterd_brickinfo_new_from_brick (brick, &newbrickinfo, - _gf_true); + _gf_true, NULL); if (ret) goto out; is_allocated = _gf_true; @@ -10266,7 +10293,7 @@ gd_should_i_start_rebalance (glusterd_volinfo_t *volinfo) { ret = glusterd_volume_brickinfo_get_by_brick (brickname, volinfo, &brick, - _gf_true); + _gf_false); if (ret) goto out; if (gf_uuid_compare (MY_UUID, brick->uuid) == 0) { diff --git a/xlators/mgmt/glusterd/src/glusterd-utils.h b/xlators/mgmt/glusterd/src/glusterd-utils.h index 9f3c4a7098a..e6436eff419 100644 --- a/xlators/mgmt/glusterd/src/glusterd-utils.h +++ b/xlators/mgmt/glusterd/src/glusterd-utils.h @@ -115,7 +115,8 @@ glusterd_brickinfo_new (glusterd_brickinfo_t **brickinfo); int32_t glusterd_brickinfo_new_from_brick (char *brick, glusterd_brickinfo_t **brickinfo, - gf_boolean_t construct_real_path); + gf_boolean_t construct_real_path, + char **op_errstr); int32_t glusterd_volinfo_find (char *volname, glusterd_volinfo_t **volinfo); diff --git a/xlators/mgmt/glusterd/src/glusterd-volgen.c b/xlators/mgmt/glusterd/src/glusterd-volgen.c index e88116e02ba..4321ddb7ddb 100644 --- a/xlators/mgmt/glusterd/src/glusterd-volgen.c +++ b/xlators/mgmt/glusterd/src/glusterd-volgen.c @@ -5116,7 +5116,8 @@ glusterd_is_valid_volfpath (char *volname, char *brick) this = THIS; GF_ASSERT (this); - ret = glusterd_brickinfo_new_from_brick (brick, &brickinfo, _gf_true); + ret = glusterd_brickinfo_new_from_brick (brick, &brickinfo, _gf_false, + NULL); if (ret) { gf_msg (this->name, GF_LOG_WARNING, 0, GD_MSG_BRICKINFO_CREATE_FAIL, diff --git a/xlators/mgmt/glusterd/src/glusterd-volume-ops.c b/xlators/mgmt/glusterd/src/glusterd-volume-ops.c index dabbdfc2e9f..7b833a9a737 100644 --- a/xlators/mgmt/glusterd/src/glusterd-volume-ops.c +++ b/xlators/mgmt/glusterd/src/glusterd-volume-ops.c @@ -1248,7 +1248,7 @@ glusterd_op_stage_create_volume (dict_t *dict, char **op_errstr, } ret = glusterd_brickinfo_new_from_brick (brick, &brick_info, - _gf_true); + _gf_true, op_errstr); if (ret) goto out; @@ -2335,7 +2335,7 @@ glusterd_op_create_volume (dict_t *dict, char **op_errstr) goto out; while ( i <= count) { ret = glusterd_brickinfo_new_from_brick (brick, &brickinfo, - _gf_true); + _gf_true, op_errstr); if (ret) goto out; -- cgit