From: Tom Lane <tgl@sss.pgh.pa.us>
To: Andres Freund <andres@anarazel.de>
cc: pgsql-hackers@lists.postgresql.org
Subject: Crash in new pgstats code
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-ID: <1656445.1650043715.1@sss.pgh.pa.us>
Content-Transfer-Encoding: quoted-printable
Date: Fri, 15 Apr 2022 13:28:35 -0400
Message-ID: <1656446.1650043715@sss.pgh.pa.us>
Archived-At: <https://www.postgresql.org/message-id/1656446.1650043715%40sss.pgh.pa.us>
Precedence: bulk

mylodon just showed a new-to-me failure mode [1]:

Core was generated by `postgres: cascade: startup recovering 0000000100000=
00000000002                '.
Program terminated with signal SIGABRT, Aborted.
#0  __GI_raise (sig=3Dsig@entry=3D6) at ../sysdeps/unix/sysv/linux/raise.c=
:49
49	../sysdeps/unix/sysv/linux/raise.c: No such file or directory.
#0  __GI_raise (sig=3Dsig@entry=3D6) at ../sysdeps/unix/sysv/linux/raise.c=
:49
#1  0x00007f8b8db2d546 in __GI_abort () at abort.c:79
#2  0x000000000098a4dc in ExceptionalCondition (conditionName=3D<optimized=
 out>, errorType=3D0x9e8061 "FailedAssertion", fileName=3D0xaf811f "/mnt/r=
esource/bf/build/mylodon/HEAD/pgsql.build/../pgsql/src/backend/lib/dshash.=
c", lineNumber=3DlineNumber@entry=3D744) at /mnt/resource/bf/build/mylodon=
/HEAD/pgsql.build/../pgsql/src/backend/utils/error/assert.c:69
#3  0x00000000006dbe65 in dshash_delete_current (status=3Dstatus@entry=3D0=
x7fffec732dc8) at /mnt/resource/bf/build/mylodon/HEAD/pgsql.build/../pgsql=
/src/backend/lib/dshash.c:744
#4  0x000000000085f911 in pgstat_free_entry (shent=3D0x7f8b8b0fc320, hstat=
=3D0x7fffec732dc8) at /mnt/resource/bf/build/mylodon/HEAD/pgsql.build/../p=
gsql/src/backend/utils/activity/pgstat_shmem.c:741
#5  pgstat_drop_entry_internal (shent=3D0x7f8b8b0fc320, hstat=3Dhstat@entr=
y=3D0x7fffec732dc8) at /mnt/resource/bf/build/mylodon/HEAD/pgsql.build/../=
pgsql/src/backend/utils/activity/pgstat_shmem.c:773
#6  0x000000000085fa2e in pgstat_drop_all_entries () at /mnt/resource/bf/b=
uild/mylodon/HEAD/pgsql.build/../pgsql/src/backend/utils/activity/pgstat_s=
hmem.c:887
#7  0x0000000000859301 in pgstat_reset_after_failure () at /mnt/resource/b=
f/build/mylodon/HEAD/pgsql.build/../pgsql/src/backend/utils/activity/pgsta=
t.c:1631
#8  pgstat_discard_stats () at /mnt/resource/bf/build/mylodon/HEAD/pgsql.b=
uild/../pgsql/src/backend/utils/activity/pgstat.c:435
#9  0x0000000000555ae0 in StartupXLOG () at /mnt/resource/bf/build/mylodon=
/HEAD/pgsql.build/../pgsql/src/backend/access/transam/xlog.c:5127
#10 0x00000000007a8ece in StartupProcessMain () at /mnt/resource/bf/build/=
mylodon/HEAD/pgsql.build/../pgsql/src/backend/postmaster/startup.c:267
#11 0x000000000079f44e in AuxiliaryProcessMain (auxtype=3Dauxtype@entry=3D=
StartupProcess) at /mnt/resource/bf/build/mylodon/HEAD/pgsql.build/../pgsq=
l/src/backend/postmaster/auxprocess.c:141
#12 0x00000000007a5891 in StartChildProcess (type=3Dtype@entry=3DStartupPr=
ocess) at /mnt/resource/bf/build/mylodon/HEAD/pgsql.build/../pgsql/src/bac=
kend/postmaster/postmaster.c:5417
#13 0x00000000007a3ea0 in PostmasterMain (argc=3Dargc@entry=3D4, argv=3D<o=
ptimized out>, argv@entry=3D0x1d1cc90) at /mnt/resource/bf/build/mylodon/H=
EAD/pgsql.build/../pgsql/src/backend/postmaster/postmaster.c:1457
#14 0x00000000006f1b91 in main (argc=3D4, argv=3D0x1d1cc90) at /mnt/resour=
ce/bf/build/mylodon/HEAD/pgsql.build/../pgsql/src/backend/main/main.c:202
$1 =3D {si_signo =3D 6, si_errno =3D 0, si_code =3D -6, _sifields =3D {_pa=
d =3D {675836, 1001, 0 <repeats 26 times>}, _kill =3D {si_pid =3D 675836, =
si_uid =3D 1001}, _timer =3D {si_tid =3D 675836, si_overrun =3D 1001, si_s=
igval =3D {sival_int =3D 0, sival_ptr =3D 0x0}}, _rt =3D {si_pid =3D 67583=
6, si_uid =3D 1001, si_sigval =3D {sival_int =3D 0, sival_ptr =3D 0x0}}, _=
sigchld =3D {si_pid =3D 675836, si_uid =3D 1001, si_status =3D 0, si_utime=
 =3D 0, si_stime =3D 0}, _sigfault =3D {si_addr =3D 0x3e9000a4ffc, _addr_l=
sb =3D 0, _addr_bnd =3D {_lower =3D 0x0, _upper =3D 0x0}}, _sigpoll =3D {s=
i_band =3D 4299262939132, si_fd =3D 0}}}

Probably deserves investigation.

			regards, tom lane

[1] https://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=3Dmylodon&dt=3D=
2022-04-15%2011%3A51%3A35