From: Patrick Donnelly <pdonnell@redhat.com>
Date: Wed, 6 Nov 2019 08:26:51 +0000 (-0800)
Subject: mds: do not try to diagnose cause of MDSMap removal
X-Git-Tag: v15.1.0~980^2~2
X-Git-Url: http://git-server-git.apps.pok.os.sepia.ceph.com/?a=commitdiff_plain;h=c385178b0c04d09bc6774acf5bdb4537fa6803ad;p=ceph.git

mds: do not try to diagnose cause of MDSMap removal

The Monitors send an empty MDSMap to an MDS it is removing. The MDS
can't diagnose the cause. Instead suggest looking at the cluster/monitor
logs.

Signed-off-by: Patrick Donnelly <pdonnell@redhat.com>
---

diff --git a/src/mds/MDSDaemon.cc b/src/mds/MDSDaemon.cc
index 6e32a588d5bc..6e65323ce027 100644
--- a/src/mds/MDSDaemon.cc
+++ b/src/mds/MDSDaemon.cc
@@ -794,25 +794,8 @@ void MDSDaemon::handle_mds_map(const cref_t<MMDSMap> &m)
   using DS = MDSMap::DaemonState;
   if (old_state != DS::STATE_NULL && new_state == DS::STATE_NULL) {
     const auto& oldinfo = oldmap->get_info_gid(mygid);
-    const auto existing = mdsmap->find_mds_gid_by_name(name);
-    if (g_conf()->mds_enforce_unique_name && existing != MDS_GID_NONE) {
-      const auto& info = mdsmap->get_info_gid(existing);
-      dout(1) << "Map replaced me " << oldinfo
-              << " with another MDS of the same name: " << info
-              << "; quitting!" << dendl;
-
-      /* Monitors should never go backwards! */
-      ceph_assert(info.global_id > myid);
-
-      // Call suicide() rather than respawn() because if someone else has
-      // taken our ID, we don't want to keep restarting and fighting them for
-      // the ID.
-      suicide();
-      return;
-    }
-
     dout(1) << "Map removed me " << oldinfo
-            << " from cluster due to lost contact; respawning" << dendl;
+            << " from cluster; respawning! See cluster/monitor logs for details." << dendl;
     respawn();
   }