Обсуждение: Spinlock error

Поиск
Список
Период
Сортировка

Spinlock error

От
"Jeff MacDonald "
Дата:
Hi folks,

I got a problem with a production database, it's a little
busy with 2 to 5 queries per second. 

Running on Redhat 6.0, 2.2.5-15 SMP (intel)

Currently thre are 128 backends ( postmaster -N 128 )
with share mem set at 256.

in pg_options verbose = 2, syslog = 2,

here are the errors..

FATAL: s_lock(4001404d) at spin.c:125, stuck spinlock. Aborting.

After those spinlock errors the postmaster died.... I tries to start it
back up and it failed with this error.

FindExec: found "/usr/bin/postgres" using argv[0]
IpcMemoryCreate: shmget failed (Identifier removed) key=5432010,
size=120, permission=700
IpcMemoryIdGet: shmget failed (Identifier removed) key=5432010, size=120,
permission=0
IpcMemoryAttach: shmat failed (Invalid argument) id=-2
FATAL 1:  AttachSLockMemory: could not attach segment
proc_exit(0) [#0]
shmem_exit(0) [#0]
exit(0)


Since this is a production database, what are teh first steps i should
take ?



======================================================
Jeff MacDonaldjeff@pgsql.com    irc: bignose on EFnet
======================================================



Re: [HACKERS] Spinlock error

От
Tom Lane
Дата:
"Jeff MacDonald <jeff@pgsql.com>" <jeffm@pgsql.com> writes:
> After those spinlock errors the postmaster died.... I tries to start it
> back up and it failed with this error.

You may need to clean up shared memory blocks and/or semaphores ---
those are supposed to be deallocated when the postmaster quits,
but it sounds like it didn't happen.  See the pgsql ipcclean script.

Not sure what caused the stuck spinlock in the first place.
        regards, tom lane


Re: [HACKERS] Spinlock error

От
"Jeff MacDonald "
Дата:
Hi Tom,

Thanks for that info, do know if thre is anyway, to
get more information from postmaster or postges ?

i'd really like to be able to solve this, i'll see if i 
can reproduce the condition.



======================================================
Jeff MacDonaldjeff@pgsql.com    irc: bignose on EFnet
======================================================

On Thu, 27 Jan 2000, Tom Lane wrote:

> "Jeff MacDonald <jeff@pgsql.com>" <jeffm@pgsql.com> writes:
> > After those spinlock errors the postmaster died.... I tries to start it
> > back up and it failed with this error.
> 
> You may need to clean up shared memory blocks and/or semaphores ---
> those are supposed to be deallocated when the postmaster quits,
> but it sounds like it didn't happen.  See the pgsql ipcclean script.
> 
> Not sure what caused the stuck spinlock in the first place.
> 
>             regards, tom lane
> 
> ************
>