From: Jianpeng Ma <jianpeng.ma@intel.com>
Date: Tue, 14 Apr 2015 01:11:58 +0000 (+0800)
Subject: osd: Fix ec pg repair endless when met unrecover object.
X-Git-Tag: v9.0.1~90^2~3
X-Git-Url: http://git-server-git.apps.pok.os.sepia.ceph.com/?a=commitdiff_plain;h=d51806f5b330d5f112281fbb95ea6addf994324e;p=ceph.git

osd: Fix ec pg repair endless when met unrecover object.

In repair_object, if bad_peer is replica, it don't add soid in
MissingLoc for ec pool. If there are more bad replica for ec pool
which cause object can't recover, the later recoverying will endless.

Signed-off-by: Jianpeng Ma <jianpeng.ma@intel.com>
Signed-off-by: Kefu Chai <kchai@redhat.com>
---

diff --git a/src/osd/PG.cc b/src/osd/PG.cc
index 03bf405ea031..7c4c6975b12f 100644
--- a/src/osd/PG.cc
+++ b/src/osd/PG.cc
@@ -3667,15 +3667,20 @@ void PG::repair_object(
     assert(waiting_for_unreadable_object.empty());
 
     pg_log.missing_add(soid, oi.version, eversion_t());
+
+    pg_log.set_last_requested(0);
+    dout(10) << __func__ << ": primary = " << primary << dendl;
+  }
+
+  if (is_ec_pg() || bad_peer == primary) {
+    // we'd better collect all shard for EC pg, and prepare good peers as the
+    // source of pull in the case of replicated pg.
     missing_loc.add_missing(soid, oi.version, eversion_t());
     list<pair<ScrubMap::object, pg_shard_t> >::iterator i;
     for (i = ok_peers->begin();
-	 i != ok_peers->end();
-	 ++i)
+	i != ok_peers->end();
+	++i)
       missing_loc.add_location(soid, i->second);
-
-    pg_log.set_last_requested(0);
-    dout(10) << __func__ << ": primary = " << primary << dendl;
   }
 }