postgres 7.0.3 hangs - Mailing list pgsql-general

This is just an addendum to my previous posting.

Here's a stack trace of the hung process:

#0  0x401887fe in select () from /lib/libc.so.6
#1  0x401e6d60 in ?? ()
#2  0x80e11a0 in LockBuffer ()
#3  0x806f862 in RelationPutHeapTupleAtEnd ()
#4  0x806ed7c in heap_insert ()
#5  0x809c3e1 in ExecAppend ()
#6  0x809c2aa in ExecutePlan ()
#7  0x809b967 in ExecutorRun ()
#8  0x80eb9aa in ProcessQueryDesc ()
#9  0x80eba10 in ProcessQuery ()
#10 0x80ea68f in pg_exec_query_dest ()
#11 0x80ea56f in pg_exec_query ()
#12 0x80eb414 in PostgresMain ()
#13 0x80d5a3a in DoBackend ()
#14 0x80d561a in BackendStartup ()
#15 0x80d49ce in ServerLoop ()
#16 0x80d445c in PostmasterMain ()
#17 0x80ab1f5 in main ()
#18 0x400fa577 in __libc_start_main () from /lib/libc.so.6


And of one of the waiting processes:

#0  0x4018dee0 in semop () from /lib/libc.so.6
#1  0x80e2ddf in IpcSemaphoreLock ()
#2  0x80e78d0 in ProcSleep ()
#3  0x80e655a in WaitOnLock ()
#4  0x80e624e in LockAcquire ()
#5  0x80e5ae7 in XactLockTableWait ()
#6  0x806f3be in heap_mark4update ()
#7  0x809c195 in ExecutePlan ()
#8  0x809b967 in ExecutorRun ()
#9  0x80a5322 in _SPI_pquery ()
#10 0x80a51b8 in _SPI_execute_plan ()
#11 0x80a4294 in SPI_execp ()
#12 0x81136b2 in RI_FKey_check ()
#13 0x811371f in RI_FKey_check_ins ()
#14 0x8095d54 in ExecCallTriggerFunc ()
#15 0x80964de in deferredTriggerExecute ()
#16 0x80965bf in deferredTriggerInvokeEvents ()
#17 0x80966db in DeferredTriggerEndQuery ()
#18 0x80eb419 in PostgresMain ()
#19 0x80d5a3a in DoBackend ()
#20 0x80d561a in BackendStartup ()
#21 0x80d49ce in ServerLoop ()
#22 0x80d445c in PostmasterMain ()
#23 0x80ab1f5 in main ()
#24 0x400fa577 in __libc_start_main () from /lib/libc.so.6

Sheer

---------- Forwarded message ----------
Date: Fri, 4 Jan 2002 12:31:02 -0500 (EST)
From: Sheer El-Showk <sheer@saraf.com>
To: pgsql-general@postgresql.org
Subject: postgres 7.0.3 hangs

Hi,

Is there any known problem with postgres 7.0.3 on a linux box (2.4 kernel)
hanging?  I noticed it from time to time in the past, but it was
infrequent enough that I didn't do anything about it, but its become a
much more pervasive problem.  I have multiple threads runing and one of
them hangs on an INSERT (visible in the process listing).  When I check
the process status, it is sleeping.  Other threads, presumably trying to
access a locked resource also hang (though they are otherwise response,
you can do other things so long as you don't access whatever resource is
locked).  Even if a I kill the sleeping process, the next one that comes
in place after it (presumably waiting on that lock) blocks in the same
way.  I have to restart the whole database to get things to work again.
I know this is not a very detailed analysis, but has anyone seen anything
like it before?  Is upgrading postgres likeley to solve the problem?

Thanks,
Sheer