From: Sage Weil <sage@redhat.com>
Date: Sat, 21 Oct 2017 03:32:33 +0000 (-0500)
Subject: messages/MOSDMap: do compat reencode of crush map, too
X-Git-Tag: v13.0.1~479^2
X-Git-Url: http://git-server-git.apps.pok.os.sepia.ceph.com/?a=commitdiff_plain;h=9eaafc66278d59447b29ce6cc378d0cd3ad14ad4;p=ceph.git

messages/MOSDMap: do compat reencode of crush map, too

If we are reencoding an incremental, and it embeds a crush map, we need
to reencode that in a compatible way too.  This is especially true now
because we have the compat crush weight-sets.  Otherwise, a client may
learn the crush map through an incremental but not understand choose_args,
and not see/understand the alternate weight set.  It will send requests
to the wrong OSDs where they will just get dropped.

Fixes: http://tracker.ceph.com/issues/21882
Signed-off-by: Sage Weil <sage@redhat.com>
---

diff --git a/src/messages/MOSDMap.h b/src/messages/MOSDMap.h
index fa46189bf050..865642cf417c 100644
--- a/src/messages/MOSDMap.h
+++ b/src/messages/MOSDMap.h
@@ -113,6 +113,14 @@ public:
 	  inc.fullmap.clear();
 	  m.encode(inc.fullmap, features | CEPH_FEATURE_RESERVED);
 	}
+	if (inc.crush.length()) {
+	  // embedded crush map
+	  CrushWrapper c;
+	  auto p = inc.crush.begin();
+	  c.decode(p);
+	  inc.crush.clear();
+	  c.encode(inc.crush, features);
+	}
 	inc.encode(p->second, features | CEPH_FEATURE_RESERVED);
       }
       for (map<epoch_t,bufferlist>::iterator p = maps.begin();