From: Samuel Just <sam.just@inktank.com>
Date: Mon, 16 Jul 2012 20:11:24 +0000 (-0700)
Subject: PG::RecoveryState::Stray::react(LogEvt&): reset last_pg_scrub
X-Git-Tag: v0.49~10
X-Git-Url: http://git-server-git.apps.pok.os.sepia.ceph.com/?a=commitdiff_plain;h=c7fb964c077d369943bd5c066c5f99da6bd5f37c;p=ceph.git

PG::RecoveryState::Stray::react(LogEvt&): reset last_pg_scrub

We need to reset the last_pg_scrub data in the osd since we
are replacing the info.

Probably fixes #2453

In cases like 2453, we hit the following backtrace:

     0> 2012-05-19 17:24:09.113684 7fe66be3d700 -1 osd/OSD.h: In function 'void OSD::unreg_last_pg_scrub(pg_t, utime_t)' thread 7fe66be3d700 time 2012-05-19 17:24:09.095719
osd/OSD.h: 840: FAILED assert(last_scrub_pg.count(p))

 ceph version 0.46-313-g4277d4d (commit:4277d4d3378dde4264e2b8d211371569219c6e4b)
 1: (OSD::unreg_last_pg_scrub(pg_t, utime_t)+0x149) [0x641f49]
 2: (PG::proc_primary_info(ObjectStore::Transaction&, pg_info_t const&)+0x5e) [0x63383e]
 3: (PG::RecoveryState::ReplicaActive::react(PG::RecoveryState::MInfoRec const&)+0x4a) [0x633eda]
 4: (boost::statechart::detail::reaction_result boost::statechart::simple_state<PG::RecoveryState::ReplicaActive, PG::RecoveryState::Started, boost::mpl::list<mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na>, (boost::statechart::history_mode)0>::local_react_impl_non_empty::local_react_impl<boost::mpl::list3<boost::statechart::custom_reaction<PG::RecoveryState::MQuery>, boost::statechart::custom_reaction<PG::RecoveryState::MInfoRec>, boost::statechart::custom_reaction<PG::RecoveryState::MLogRec> >, boost::statechart::simple_state<PG::RecoveryState::ReplicaActive, PG::RecoveryState::Started, boost::mpl::list<mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na>, (boost::statechart::history_mode)0> >(boost::statechart::simple_state<PG::RecoveryState::ReplicaActive, PG::RecoveryState::Started, boost::mpl::list<mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na>, (boost::statechart::history_mode)0>&, boost::statechart::event_base const&, void const*)+0x130) [0x6466a0]
 5: (boost::statechart::simple_state<PG::RecoveryState::ReplicaActive, PG::RecoveryState::Started, boost::mpl::list<mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na, mpl_::na>, (boost::statechart::history_mode)0>::react_impl(boost::statechart::event_base const&, void const*)+0x81) [0x646791]
 6: (boost::statechart::state_machine<PG::RecoveryState::RecoveryMachine, PG::RecoveryState::Initial, std::allocator<void>, boost::statechart::null_exception_translator>::send_event(boost::statechart::event_base const&)+0x5b) [0x63dfcb]
 7: (boost::statechart::state_machine<PG::RecoveryState::RecoveryMachine, PG::RecoveryState::Initial, std::allocator<void>, boost::statechart::null_exception_translator>::process_event(boost::statechart::event_base const&)+0x11) [0x63e0f1]
 8: (PG::RecoveryState::handle_info(int, pg_info_t&, PG::RecoveryCtx*)+0x177) [0x616987]
 9: (OSD::handle_pg_info(std::tr1::shared_ptr<OpRequest>)+0x665) [0x5d3d15]
 10: (OSD::dispatch_op(std::tr1::shared_ptr<OpRequest>)+0x2a0) [0x5d7370]
 11: (OSD::_dispatch(Message*)+0x191) [0x5dd4a1]
 12: (OSD::ms_dispatch(Message*)+0x153) [0x5ddda3]
 13: (SimpleMessenger::dispatch_entry()+0x863) [0x77fbc3]
 14: (SimpleMessenger::DispatchThread::entry()+0xd) [0x746c5d]
 15: (()+0x7efc) [0x7fe679b1fefc]
 16: (clone()+0x6d) [0x7fe67815089d]
 NOTE: a copy of the executable, or `objdump -rdS <executable>` is needed to interpret this.

Because we don't clear the scrub state before reseting info,
the last_scrub_stamp state in the info.history structure
changes without updating the osd state resulting in the
above assert failure.

Backport: stable

Signed-off-by: Samuel Just <sam.just@inktank.com>
---

diff --git a/src/osd/PG.cc b/src/osd/PG.cc
index 14d415b6e47bd..678b43519e22e 100644
--- a/src/osd/PG.cc
+++ b/src/osd/PG.cc
@@ -4475,7 +4475,11 @@ boost::statechart::result PG::RecoveryState::Stray::react(const MLogRec& logevt)
 
   if (msg->info.last_backfill == hobject_t()) {
     // restart backfill
+    pg->osd->unreg_last_pg_scrub(pg->info.pgid,
+				 pg->info.history.last_scrub_stamp);
     pg->info = msg->info;
+    pg->osd->reg_last_pg_scrub(pg->info.pgid,
+			       pg->info.history.last_scrub_stamp);
     pg->log.claim_log(msg->log);
     pg->missing.clear();
   } else {