Bug #45795
Updated by Samuel Just almost 4 years ago
I'm running into this assert while trying to exercise krbd with replica reads (particularly balanced reads):
<pre>
-3> 2020-05-31T01:41:03.454+0000 7feef4815700 10 osd.4 pg_epoch: 160 pg[3.e(unlocked)] -- Watch(140522136622944,client.4455) HandleWatchTimeout
-2> 2020-05-31T01:41:03.454+0000 7feef4815700 10 osd.4 pg_epoch: 160 pg[3.e( v 160'3343 (0'0,160'3343] local-lis/les=19/20 n=217 ec=19/19 lis/c=19/19 les/c/f=20/20/0 sis=19) [7,4] r=1 lpr=19 luod=0'0 crt=160'3341 mlcod 159'3328 active
mbc={}] handle_watch_timeout obc 0x55d0e1f2f440
-1> 2020-05-31T01:41:03.462+0000 7feef4815700 -1 /build/ceph-16.0.0-2020-g64aa691/src/osd/PrimaryLogPG.cc: In function 'bool PrimaryLogPG::is_degraded_or_backfilling_object(const hobject_t&)' thread 7feef4815700 time 2020-05-31T01:41:
03.458493+0000
/build/ceph-16.0.0-2020-g64aa691/src/osd/PrimaryLogPG.cc: 627: FAILED ceph_assert(!get_acting_recovery_backfill().empty())
ceph version 16.0.0-2020-g64aa691 (64aa6910df80dea3468fff35edf20c3c10ff9b1b) pacific (dev)
1: (ceph::__ceph_assert_fail(char const*, char const*, int, char const*)+0x154) [0x55d0d40d2c8c]
2: (ceph::__ceph_assertf_fail(char const*, char const*, int, char const*, char const*, ...)+0) [0x55d0d40d2e67]
3: (()+0xb8b437) [0x55d0d424e437]
4: (PrimaryLogPG::handle_watch_timeout(std::shared_ptr<Watch>)+0xb8) [0x55d0d4276258]
5: (HandleWatchTimeout::complete(int)+0x118) [0x55d0d41e9618]
6: (SafeTimer::timer_thread()+0x1ac) [0x55d0d47b4bac]
7: (SafeTimerThread::entry()+0xd) [0x55d0d47b5f7d]
8: (()+0x76db) [0x7fef1801c6db]
9: (clone()+0x3f) [0x7fef16dbc88f]
0> 2020-05-31T01:41:03.466+0000 7feef4815700 -1 *** Caught signal (Aborted) **
in thread 7feef4815700 thread_name:safe_timer
</pre>
Might be similar in nature to #15391?