Ceph

3> 2015-01-22 22:53:26.560136 7f621276e700 10 osd.3 514 handle_replica_op osd_repop_reply(osd.3.0:3457 2.0 ack, result = 0) v1 epoch 514
-2> 2015-01-22 22:53:26.560143 7f621276e700 20 osd.3 514 should_share_map osd.5 10.214.131.19:6809/28426 514
-1> 2015-01-22 22:53:26.560149 7f621276e700 15 osd.3 514 enqueue_op 0x4c4c600 prio 196 cost 0 latency 0.000128 osd_repop_reply(osd.3.0:3457 2.0 ack, result = 0) v1
0> 2015-01-22 22:53:26.560643 7f621939b700 -1 osd/ReplicatedPG.cc: In function 'void ReplicatedPG::finish_ctx(ReplicatedPG::OpContext*, int, bool, bool)' thread 7f621939b700 time 2015-01-22 22:53:26.557192
osd/ReplicatedPG.cc: 5790: FAILED assert(ctx>new_obs.exists == ctx->new_snapset.head_exists)

ceph version 0.91-738-g17ecbb0 (17ecbb0a6af400cbf458cc707a99d9c5063a1697)
1: (ceph::__ceph_assert_fail(char const*, char const*, int, char const*)+0x7f) [0xae601f]
2: (ReplicatedPG::finish_ctx(ReplicatedPG::OpContext*, int, bool, bool)+0x306a) [0x834cfa]
3: (ReplicatedPG::_scrub(ScrubMap&)+0x989) [0x836f69]
4: (PG::scrub_compare_maps()+0xfbe) [0x79ee5e]
5: (PG::chunky_scrub(ThreadPool::TPHandle&)+0x1c9) [0x79f7f9]
6: (PG::scrub(ThreadPool::TPHandle&)+0x33d) [0x7a11ed]
7: (OSD::ScrubWQ::_process(PG*, ThreadPool::TPHandle&)+0x13) [0x68c753]
8: (ThreadPool::worker(ThreadPool::WorkThread*)+0x48e) [0xad563e]
9: (ThreadPool::WorkThread::entry()+0x10) [0xad83d0]
10: (()+0x7e9a) [0x7f62380c2e9a]
11: (clone()+0x6d) [0x7f623686c31d]
NOTE: a copy of the executable, or `objdump -rdS <executable>` is needed to interpret this.

--- logging levels ---

hit_set_persist updates new_snapset but not ssc->snapset. Also, we might as well write down the digest during hit_set_persist.