glusterd: cleanup pidfile on pmap signout

This patch ensures
1. brick pidfile is cleaned up on pmap signout
2. pmap signout evemt is sent for all the bricks when a brick process
shuts down.

>Reviewed-on: https://review.gluster.org/17168
>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: Jeff Darcy <jeff@pl.atyp.us>
>(cherry picked from commit 3d35e21ffb15713237116d85711e9cd1dda1688a)

Change-Id: I7606a60775b484651d4b9743b6037b40323931a2
BUG: 1449002
Signed-off-by: Atin Mukherjee <amukherj@redhat.com>
Reviewed-on: https://review.gluster.org/17209
Smoke: Gluster Build System <jenkins@build.gluster.org>
Reviewed-by: Prashanth Pai <ppai@redhat.com>
NetBSD-regression: NetBSD Build System <jenkins@build.gluster.org>
CentOS-regression: Gluster Build System <jenkins@build.gluster.org>
Reviewed-by: Jeff Darcy <jeff@pl.atyp.us>

4 files changed, 109 insertions, 6 deletions

diff --git a/glusterfsd/src/glusterfsd.c b/glusterfsd/src/glusterfsd.c
index 41515fe12e7..dd89c47ff43 100644
--- a/glusterfsd/src/glusterfsd.c
+++ b/glusterfsd/src/glusterfsd.c
@@ -1303,6 +1303,9 @@ cleanup_and_exit (int signum)
 {
         glusterfs_ctx_t *ctx      = NULL;
         xlator_t        *trav     = NULL;
+        xlator_t        *top;
+        xlator_t        *victim;
+        xlator_list_t   **trav_p;
 
         ctx = glusterfsd_ctx;
 
@@ -1332,7 +1335,20 @@ cleanup_and_exit (int signum)
                 return;
 
         ctx->cleanup_started = 1;
-        glusterfs_mgmt_pmap_signout (ctx, NULL);
+
+        /* signout should be sent to all the bricks in case brick mux is enabled
+         * and multiple brick instances are attached to this process
+         */
+        if (ctx->active) {
+                top = ctx->active->first;
+                for (trav_p = &top->children; *trav_p;
+                     trav_p = &(*trav_p)->next) {
+                        victim = (*trav_p)->xlator;
+                        glusterfs_mgmt_pmap_signout (ctx, victim->name);
+                }
+        } else {
+                glusterfs_mgmt_pmap_signout (ctx, NULL);
+        }
 
         /* below part is a racy code where the rpcsvc object is freed.
          * But in another thread (epoll thread), upon poll error in the
diff --git a/xlators/mgmt/glusterd/src/glusterd-pmap.c b/xlators/mgmt/glusterd/src/glusterd-pmap.c
index 9b2954af64d..0507715305c 100644
--- a/xlators/mgmt/glusterd/src/glusterd-pmap.c
+++ b/xlators/mgmt/glusterd/src/glusterd-pmap.c
@@ -519,11 +519,20 @@ gluster_pmap_signin (rpcsvc_request_t *req)
 int
 __gluster_pmap_signout (rpcsvc_request_t *req)
 {
-        pmap_signout_req    args                 = {0,};
-        pmap_signout_rsp    rsp                  = {0,};
-        int                 ret                  = -1;
-        char                brick_path[PATH_MAX] = {0,};
-        glusterd_brickinfo_t *brickinfo = NULL;
+        pmap_signout_req      args                 = {0,};
+        pmap_signout_rsp      rsp                  = {0,};
+        int                   ret                  = -1;
+        xlator_t             *this                 = NULL;
+        glusterd_conf_t      *conf                 = NULL;
+        glusterd_volinfo_t   *volinfo              = NULL;
+        glusterd_brickinfo_t *brickinfo            = NULL;
+        char                  pidfile[PATH_MAX]    = {0};
+        char                  brick_path[PATH_MAX] = {0,};
+
+        this = THIS;
+        GF_VALIDATE_OR_GOTO ("glusterd", this, fail);
+        conf = this->private;
+        GF_VALIDATE_OR_GOTO (this->name, conf, fail);
 
         ret = xdr_to_generic (req->msg[0], &args,
                               (xdrproc_t)xdr_pmap_signout_req);
@@ -542,6 +551,18 @@ __gluster_pmap_signout (rpcsvc_request_t *req)
                                 brick_path, GF_PMAP_PORT_BRICKSERVER,
                                 req->trans);
         }
+        /* Clean up the pidfile for this brick given glusterfsd doesn't clean it
+         * any more. This is required to ensure we don't end up with having
+         * stale pid files in case a brick is killed from the backend
+         */
+        ret = glusterd_get_volinfo_from_brick (args.brick, &volinfo);
+        if (!ret) {
+                if (volinfo && brickinfo) {
+                        GLUSTERD_GET_BRICK_PIDFILE (pidfile, volinfo, brickinfo,
+                                                    conf);
+                        sys_unlink (pidfile);
+                }
+        }
 
 fail:
         glusterd_submit_reply (req, &rsp, NULL, 0, NULL,
diff --git a/xlators/mgmt/glusterd/src/glusterd-utils.c b/xlators/mgmt/glusterd/src/glusterd-utils.c
index df058330cd0..e7fd01aabb2 100644
--- a/xlators/mgmt/glusterd/src/glusterd-utils.c
+++ b/xlators/mgmt/glusterd/src/glusterd-utils.c
@@ -5827,6 +5827,7 @@ glusterd_get_brickinfo (xlator_t *this, const char *brickname, int port,
         glusterd_conf_t         *priv = NULL;
         glusterd_volinfo_t      *volinfo = NULL;
         glusterd_brickinfo_t    *tmpbrkinfo = NULL;
+        glusterd_snap_t         *snap    = NULL;
         int                     ret = -1;
 
         GF_ASSERT (brickname);
@@ -5845,6 +5846,21 @@ glusterd_get_brickinfo (xlator_t *this, const char *brickname, int port,
                         }
                 }
         }
+        /* In case normal volume is not found, check for snapshot volumes */
+        cds_list_for_each_entry (snap, &priv->snapshots, snap_list) {
+                cds_list_for_each_entry (volinfo, &snap->volumes, vol_list) {
+                        cds_list_for_each_entry (tmpbrkinfo, &volinfo->bricks,
+                                                 brick_list) {
+                                if (gf_uuid_compare (tmpbrkinfo->uuid, MY_UUID))
+                                        continue;
+                                if (!strcmp(tmpbrkinfo->path, brickname)) {
+                                        *brickinfo = tmpbrkinfo;
+                                        return 0;
+                                }
+                        }
+                }
+        }
+
         return ret;
 }
 
@@ -13212,6 +13228,52 @@ out:
         return ret;
 }
 
+int
+glusterd_get_volinfo_from_brick (char *brick, glusterd_volinfo_t **volinfo)
+{
+        int             ret                 = -1;
+        xlator_t                *this       = NULL;
+        glusterd_conf_t         *conf       = NULL;
+        glusterd_volinfo_t      *voliter    = NULL;
+        glusterd_brickinfo_t    *brickiter  = NULL;
+        glusterd_snap_t         *snap    = NULL;
+
+        this = THIS;
+        GF_VALIDATE_OR_GOTO ("glusterd", this, out);
+        conf = this->private;
+        GF_VALIDATE_OR_GOTO (this->name, conf, out);
+
+        /* First check for normal volumes */
+        cds_list_for_each_entry (voliter, &conf->volumes, vol_list) {
+                cds_list_for_each_entry (brickiter, &voliter->bricks,
+                                         brick_list) {
+                        if (gf_uuid_compare (brickiter->uuid, MY_UUID))
+                                continue;
+                        if (!strcmp(brickiter->path, brick)) {
+                                *volinfo = voliter;
+                                return 0;
+                        }
+                }
+        }
+        /* In case normal volume is not found, check for snapshot volumes */
+        cds_list_for_each_entry (snap, &conf->snapshots, snap_list) {
+                cds_list_for_each_entry (voliter, &snap->volumes, vol_list) {
+                        cds_list_for_each_entry (brickiter, &voliter->bricks,
+                                                 brick_list) {
+                                if (gf_uuid_compare (brickiter->uuid, MY_UUID))
+                                        continue;
+                                if (!strcmp(brickiter->path, brick)) {
+                                        *volinfo = voliter;
+                                        return 0;
+                                }
+                        }
+                }
+        }
+
+out:
+        return ret;
+}
+
 glusterd_op_t
 gd_cli_to_gd_op (char *cli_op)
 {
diff --git a/xlators/mgmt/glusterd/src/glusterd-utils.h b/xlators/mgmt/glusterd/src/glusterd-utils.h
index db13c4c8ad4..31bf462bcad 100644
--- a/xlators/mgmt/glusterd/src/glusterd-utils.h
+++ b/xlators/mgmt/glusterd/src/glusterd-utils.h
@@ -847,4 +847,8 @@ glusterd_brick_op_prerequisites (dict_t *dict,
                                  char **src_brick, glusterd_brickinfo_t
                                  **src_brickinfo, char *pidfile,
                                  char **op_errstr, dict_t *rsp_dict);
+
+int
+glusterd_get_volinfo_from_brick (char *brick, glusterd_volinfo_t **volinfo);
+
 #endif