feature/bitrot: Ondemand scrub option for bitrot

The bitrot scrubber takes 'hourly/daily/biweekly/monthly'
as the values for 'scrub-frequency'. There is no way
to schedule the scrubbing when the admin wants it.

Ondemand scrubbing brings in the new option 'ondemand'
with which the admin can start scrubbing ondemand.
It starts the scrubbing immediately.

Ondemand scrubbing is successful only if the scrubber
is in 'Active (Idle)' (waiting for it's next frequency
cycle to start scrubbing). It is not entertained when
the scrubber is in 'Paused' or already running.

Here is the command line syntax.

gluster volume bitrot <vol name> scrub ondemand

Change-Id: I84c28904367eed827a7dae8d6a535c14b28e9f4d
BUG: 1366195
Signed-off-by: Kotresh HR <khiremat@redhat.com>
Reviewed-on: http://review.gluster.org/15111
Smoke: Gluster Build System <jenkins@build.gluster.org>
NetBSD-regression: NetBSD Build System <jenkins@build.gluster.org>
CentOS-regression: Gluster Build System <jenkins@build.gluster.org>
Reviewed-by: Venky Shankar <vshankar@redhat.com>

5 files changed, 47 insertions, 4 deletions

diff --git a/xlators/mgmt/glusterd/src/glusterd-bitrot.c b/xlators/mgmt/glusterd/src/glusterd-bitrot.c
index 6e91106c8e5..8c5ddfd7896 100644
--- a/xlators/mgmt/glusterd/src/glusterd-bitrot.c
+++ b/xlators/mgmt/glusterd/src/glusterd-bitrot.c
@@ -138,6 +138,34 @@ __glusterd_handle_bitrot (rpcsvc_request_t *req)
                 }
         }
 
+        if (type == GF_BITROT_CMD_SCRUB_ONDEMAND) {
+                /* Backward compatibility handling for scrub status command*/
+                if (conf->op_version < GD_OP_VERSION_3_9_0) {
+                        snprintf (msg, sizeof (msg), "Cannot execute command. "
+                                  "The cluster is operating at version %d. "
+                                  "Bitrot scrub ondemand command unavailable in "
+                                  "this version", conf->op_version);
+                        ret = -1;
+                        goto out;
+                }
+
+                ret = dict_get_str (dict, "scrub-value", &scrub);
+                if (ret) {
+                        gf_msg (this->name, GF_LOG_ERROR, 0,
+                                GD_MSG_DICT_GET_FAILED,
+                                "Failed to get scrub value.");
+                        ret = -1;
+                        goto out;
+                }
+
+                if (!strncmp (scrub, "ondemand", strlen ("ondemand"))) {
+                        ret = glusterd_op_begin_synctask (req,
+                                                          GD_OP_SCRUB_ONDEMAND,
+                                                          dict);
+                        goto out;
+                }
+        }
+
         ret = glusterd_op_begin_synctask (req, GD_OP_BITROT, dict);
 
 out:
@@ -572,6 +600,7 @@ glusterd_op_bitrot (dict_t *dict, char **op_errstr, dict_t *rsp_dict)
                 if (ret)
                         goto out;
         case GF_BITROT_CMD_SCRUB_STATUS:
+        case GF_BITROT_CMD_SCRUB_ONDEMAND:
                 break;
 
         default:
diff --git a/xlators/mgmt/glusterd/src/glusterd-op-sm.c b/xlators/mgmt/glusterd/src/glusterd-op-sm.c
index bae9be872f4..b4eb8b13c66 100644
--- a/xlators/mgmt/glusterd/src/glusterd-op-sm.c
+++ b/xlators/mgmt/glusterd/src/glusterd-op-sm.c
@@ -714,6 +714,7 @@ glusterd_node_op_build_payload (glusterd_op_t op, gd1_mgmt_brick_op_req **req,
                 break;
 
         case GD_OP_SCRUB_STATUS:
+        case GD_OP_SCRUB_ONDEMAND:
                 brick_req = GF_CALLOC (1, sizeof(*brick_req),
                                        gf_gld_mt_mop_brick_req_t);
                 if (!brick_req)
@@ -4131,6 +4132,7 @@ glusterd_op_build_payload (dict_t **req, char **op_errstr, dict_t *op_ctx)
                 case GD_OP_BARRIER:
                 case GD_OP_BITROT:
                 case GD_OP_SCRUB_STATUS:
+                case GD_OP_SCRUB_ONDEMAND:
                         {
                                 do_common = _gf_true;
                         }
@@ -4725,6 +4727,7 @@ glusterd_op_modify_op_ctx (glusterd_op_t op, void *ctx)
          */
         case GD_OP_DEFRAG_BRICK_VOLUME:
         case GD_OP_SCRUB_STATUS:
+        case GD_OP_SCRUB_ONDEMAND:
                 ret = dict_get_int32 (op_ctx, "count", &count);
                 if (ret) {
                         gf_msg_debug (this->name, 0,
@@ -4772,10 +4775,11 @@ glusterd_op_modify_op_ctx (glusterd_op_t op, void *ctx)
                                 GD_MSG_CONVERSION_FAILED,
                                 "Failed uuid to hostname conversion");
 
-                /* Since Both rebalance and bitrot scrub status are going to
-                 * use same code path till here, we should break in case
-                 * of scrub status */
-                if (op == GD_OP_SCRUB_STATUS) {
+                /* Since Both rebalance and bitrot scrub status/ondemand
+                 * are going to use same code path till here, we should
+                 * break in case of scrub status.
+                 */
+                if (op == GD_OP_SCRUB_STATUS || op == GD_OP_SCRUB_ONDEMAND) {
                         break;
                 }
 
@@ -5442,6 +5446,7 @@ glusterd_need_brick_op (glusterd_op_t op)
         case GD_OP_DEFRAG_BRICK_VOLUME:
         case GD_OP_HEAL_VOLUME:
         case GD_OP_SCRUB_STATUS:
+        case GD_OP_SCRUB_ONDEMAND:
                 ret = _gf_true;
                 break;
         default:
@@ -5713,6 +5718,7 @@ glusterd_op_stage_validate (glusterd_op_t op, dict_t *dict, char **op_errstr,
 
                 case GD_OP_BITROT:
                 case GD_OP_SCRUB_STATUS:
+                case GD_OP_SCRUB_ONDEMAND:
                         ret = glusterd_op_stage_bitrot (dict, op_errstr,
                                                         rsp_dict);
                         break;
@@ -5838,6 +5844,7 @@ glusterd_op_commit_perform (glusterd_op_t op, dict_t *dict, char **op_errstr,
 
                 case GD_OP_BITROT:
                 case GD_OP_SCRUB_STATUS:
+                case GD_OP_SCRUB_ONDEMAND:
                         ret = glusterd_op_bitrot (dict, op_errstr, rsp_dict);
                         break;
 
@@ -7288,6 +7295,7 @@ glusterd_op_bricks_select (glusterd_op_t op, dict_t *dict, char **op_errstr,
                 ret = glusterd_bricks_select_snap (dict, op_errstr, selected);
                 break;
         case GD_OP_SCRUB_STATUS:
+        case GD_OP_SCRUB_ONDEMAND:
                 ret = glusterd_bricks_select_scrub (dict, op_errstr, selected);
                 break;
         default:
diff --git a/xlators/mgmt/glusterd/src/glusterd-rpc-ops.c b/xlators/mgmt/glusterd/src/glusterd-rpc-ops.c
index 890ccf06cdc..9cf6fda5152 100644
--- a/xlators/mgmt/glusterd/src/glusterd-rpc-ops.c
+++ b/xlators/mgmt/glusterd/src/glusterd-rpc-ops.c
@@ -143,6 +143,7 @@ glusterd_op_send_cli_response (glusterd_op_t op, int32_t op_ret,
         case GD_OP_BARRIER:
         case GD_OP_BITROT:
         case GD_OP_SCRUB_STATUS:
+        case GD_OP_SCRUB_ONDEMAND:
         {
                 /*nothing specific to be done*/
                 break;
diff --git a/xlators/mgmt/glusterd/src/glusterd-syncop.c b/xlators/mgmt/glusterd/src/glusterd-syncop.c
index 7c5721f25d0..7aeda77a59d 100644
--- a/xlators/mgmt/glusterd/src/glusterd-syncop.c
+++ b/xlators/mgmt/glusterd/src/glusterd-syncop.c
@@ -309,6 +309,10 @@ glusterd_syncop_aggr_rsp_dict (glusterd_op_t op, dict_t *aggr, dict_t *rsp)
         case GD_OP_SCRUB_STATUS:
                 ret = glusterd_volume_bitrot_scrub_use_rsp_dict (aggr, rsp);
         break;
+
+        case GD_OP_SCRUB_ONDEMAND:
+        break;
+
         default:
         break;
         }
diff --git a/xlators/mgmt/glusterd/src/glusterd.h b/xlators/mgmt/glusterd/src/glusterd.h
index 77eded4f29c..c07fe42fe61 100644
--- a/xlators/mgmt/glusterd/src/glusterd.h
+++ b/xlators/mgmt/glusterd/src/glusterd.h
@@ -117,6 +117,7 @@ typedef enum glusterd_op_ {
         GD_OP_DETACH_TIER,
         GD_OP_TIER_MIGRATE,
         GD_OP_SCRUB_STATUS,
+        GD_OP_SCRUB_ONDEMAND,
         GD_OP_MAX,
 } glusterd_op_t;