glusterd/snapshot-handshake: Perform handshake of missed_snaps_list.

In a handshake, create a union of the missed_snap_lists of the two peers.
If an entry is present, its no op.
If an entry is pendng, and the peer entry is done, mark own entry as done.
If an entry is done, and the peer ertry is pending, its a no-op.
If its a new entry, add it.

Change-Id: Idbfa49cc34871631ba8c7c56d915666311024887
BUG: 1061685
Signed-off-by: Avra Sengupta <asengupt@redhat.com>
Reviewed-on: http://review.gluster.org/7453
Tested-by: Gluster Build System <jenkins@build.gluster.com>
Reviewed-by: Rajesh Joseph <rjoseph@redhat.com>
Reviewed-by: Vijay Bellur <vbellur@redhat.com>

6 files changed, 151 insertions, 0 deletions

diff --git a/xlators/mgmt/glusterd/src/glusterd-handler.c b/xlators/mgmt/glusterd/src/glusterd-handler.c
index 53c402136dd..5869a88d4be 100644
--- a/xlators/mgmt/glusterd/src/glusterd-handler.c
+++ b/xlators/mgmt/glusterd/src/glusterd-handler.c
@@ -3403,6 +3403,12 @@ 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;
+
                         default:
                                 snprintf (errstr, len, "Probe returned with "
                                           "unknown errno %d", op_errno);
diff --git a/xlators/mgmt/glusterd/src/glusterd-rpc-ops.c b/xlators/mgmt/glusterd/src/glusterd-rpc-ops.c
index 27910d13206..cf1ae279a00 100644
--- a/xlators/mgmt/glusterd/src/glusterd-rpc-ops.c
+++ b/xlators/mgmt/glusterd/src/glusterd-rpc-ops.c
@@ -1266,6 +1266,16 @@ glusterd_rpc_friend_add (call_frame_t *frame, xlator_t *this,
         if (ret)
                 goto out;
 
+        if (priv->op_version >= GD_OP_VERSION_4) {
+                ret = glusterd_add_missed_snaps_to_export_dict (vols);
+                if (ret) {
+                        gf_log (this->name, GF_LOG_ERROR,
+                                "Unable to add list of missed snapshots "
+                                "in the vols dict for handshake");
+                        goto out;
+                }
+        }
+
         uuid_copy (req.uuid, MY_UUID);
         req.hostname = peerinfo->hostname;
         req.port = peerinfo->port;
diff --git a/xlators/mgmt/glusterd/src/glusterd-sm.c b/xlators/mgmt/glusterd/src/glusterd-sm.c
index fd56e5abfcc..3f0add22056 100644
--- a/xlators/mgmt/glusterd/src/glusterd-sm.c
+++ b/xlators/mgmt/glusterd/src/glusterd-sm.c
@@ -648,10 +648,14 @@ 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;
 
+        this = THIS;
+        GF_ASSERT (this);
         GF_ASSERT (ctx);
         ev_ctx = ctx;
         uuid_copy (uuid, ev_ctx->uuid);
@@ -659,6 +663,9 @@ 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);
@@ -674,6 +681,21 @@ 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_friend_sm_new_event (event_type, &new_event);
 
         if (ret) {
diff --git a/xlators/mgmt/glusterd/src/glusterd-utils.c b/xlators/mgmt/glusterd/src/glusterd-utils.c
index be2022da42d..74317eb446d 100644
--- a/xlators/mgmt/glusterd/src/glusterd-utils.c
+++ b/xlators/mgmt/glusterd/src/glusterd-utils.c
@@ -2667,6 +2667,64 @@ out:
 }
 
 int32_t
+glusterd_add_missed_snaps_to_export_dict (dict_t *vols)
+{
+        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 (vols);
+
+        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=%d:%s:%d:%d",
+                                  missed_snapinfo->node_snap_info,
+                                  snap_opinfo->brick_num,
+                                  snap_opinfo->brick_path,
+                                  snap_opinfo->op,
+                                  snap_opinfo->status);
+
+                        ret = dict_set_dynstr_with_alloc (vols, 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 (vols, "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_build_volume_dict (dict_t **vols)
 {
         int32_t                 ret = -1;
@@ -4156,6 +4214,44 @@ out:
         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 *vols)
+{
+        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 (vols);
+
+        priv = this->private;
+        GF_ASSERT (priv);
+
+        /* Add the friends missed_snaps entries to the in-memory list */
+        ret = dict_get_int32 (vols, "missed_snap_count", &missed_snap_count);
+        if (ret) {
+                gf_log (this->name, GF_LOG_INFO,
+                        "No missed snaps");
+                ret = 0;
+                goto out;
+        }
+
+        ret = glusterd_store_update_missed_snaps (vols, missed_snap_count);
+        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;
+}
+
 int32_t
 glusterd_compare_friend_data (dict_t  *vols, int32_t *status, char *hostname)
 {
diff --git a/xlators/mgmt/glusterd/src/glusterd-utils.h b/xlators/mgmt/glusterd/src/glusterd-utils.h
index d2121778a10..0f1c5237294 100644
--- a/xlators/mgmt/glusterd/src/glusterd-utils.h
+++ b/xlators/mgmt/glusterd/src/glusterd-utils.h
@@ -719,4 +719,10 @@ glusterd_add_missed_snaps_to_dict (dict_t *rsp_dict, char *snap_uuid,
                                    glusterd_brickinfo_t *brickinfo,
                                    int32_t brick_number, int32_t op);
 
+int32_t
+glusterd_add_missed_snaps_to_export_dict (dict_t *vols);
+
+int32_t
+glusterd_import_friend_missed_snap_list (dict_t *vols);
+
 #endif
diff --git a/xlators/mgmt/glusterd/src/glusterd.c b/xlators/mgmt/glusterd/src/glusterd.c
index abef538d8f0..1a6aa81d30c 100644
--- a/xlators/mgmt/glusterd/src/glusterd.c
+++ b/xlators/mgmt/glusterd/src/glusterd.c
@@ -1256,6 +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);