From: Samuel Just <sjust@redhat.com>
Date: Mon, 17 Aug 2015 18:21:27 +0000 (-0700)
Subject: ReplicatedPG: treat object as degraded until on_global_recover
X-Git-Tag: v9.1.0~345^2~1
X-Git-Url: http://git-server-git.apps.pok.os.sepia.ceph.com/?a=commitdiff_plain;h=79f310ac4d3d2c520e7665be057f2143113537a9;p=ceph.git

ReplicatedPG: treat object as degraded until on_global_recover

Otherwise, an op might slip by between on_local_recover and
on_global_recover.

Signed-off-by: Samuel Just <sjust@redhat.com>
---

diff --git a/src/osd/ReplicatedPG.cc b/src/osd/ReplicatedPG.cc
index 48f4277dbda1..98078f85d1f0 100644
--- a/src/osd/ReplicatedPG.cc
+++ b/src/osd/ReplicatedPG.cc
@@ -413,6 +413,12 @@ void ReplicatedPG::wait_for_all_missing(OpRequestRef op)
 
 bool ReplicatedPG::is_degraded_or_backfilling_object(const hobject_t& soid)
 {
+  /* The conditions below may clear (on_local_recover, before we queue
+   * the tranasction) before we actually requeue the degraded waiters
+   * in on_global_recover after the transaction completes.
+   */
+  if (waiting_for_degraded_object.count(soid))
+    return true;
   if (pg_log.get_missing().missing.count(soid))
     return true;
   assert(!actingbackfill.empty());