Обсуждение: postgres dumps core - HASH indexes

After a few minutes, my postgres dumps core.


(gdb) where
#0  0x041ee52d in kill () from /usr/lib/libc.so.40.3
#1  0x04226c37 in abort () at /usr/src/lib/libc/stdlib/abort.c:65
#2  0x1c1c2fcb in errfinish ()
#3  0x1c1c3be1 in elog_finish ()
#4  0x1c040598 in _hash_expandtable ()
#5  0x1c03f1a4 in _hash_doinsert ()
#6  0x1c03deee in hashinsert ()
#7  0x1c1c76e0 in FunctionCall6 ()
#8  0x1c0484b6 in index_insert ()
#9  0x1c0e7c84 in ExecInsertIndexTuples ()
#10 0x1c0e011b in ExecUpdate ()
#11 0x1c0df999 in ExecutePlan ()
#12 0x1c0deccf in ExecutorRun ()
#13 0x1c15ac05 in ProcessQuery ()
#14 0x1c15ba75 in PortalRunMulti ()
#15 0x1c15b3fa in PortalRun ()
#16 0x1c15712b in exec_simple_query ()
#17 0x1c159f8e in PostgresMain ()
#18 0x1c137c8c in BackendRun ()
#19 0x1c13755b in BackendStartup ()
#20 0x1c13594f in ServerLoop ()
#21 0x1c134f1a in PostmasterMain ()
#22 0x1c0fd3e4 in main ()
(gdb)


This is the stacktrace.

Is it possible to find out why? I use hash indexes on some tables, 
probably it's related to this problem.

Thanks.

Server/Client: 8.2.3
Host OS: OpenBSD arsenic.digitalinfluence.hu 4.1 GENERIC.MP#1 i386 Intel 
Pentium III ("GenuineIntel" 686-class)

Adam PAPAI <wooh@wooh.hu> writes:
> After a few minutes, my postgres dumps core.
> (gdb) where
> #0  0x041ee52d in kill () from /usr/lib/libc.so.40.3
> #1  0x04226c37 in abort () at /usr/src/lib/libc/stdlib/abort.c:65
> #2  0x1c1c2fcb in errfinish ()
> #3  0x1c1c3be1 in elog_finish ()
> #4  0x1c040598 in _hash_expandtable ()

> Server/Client: 8.2.3

This is probably the same bug fixed in 8.2.4:

* Fix PANIC during enlargement of a hash index (Tom)
        regards, tom lane