Thread: Seeking advice about the dreaded stuck spinlock problem on NT

Seeking advice about the dreaded stuck spinlock problem on NT

From
Dan Rosner
Date:
Does anyone have any advice/suggestions about how to deal with
stuck spinlock problems while running postgres 6.5.3
on cygwin B20?

Background:

Running on NT with sp6, I'm using active perl with DBI
and DBD-PG to load data into postgres and update some tables
every five minutes with a simple batch file that executes my
perl scripts.  Everything seems to work well for a few hours,
but then the backend dies with the dreaded...

FATAL: s_lock(14490065) at spin.c:125, stuck spinlock. Aborting.

After this happens, I'm unable to connect to the backend unless I
kill and restart the ipc-daemon and postmaster again from the cygwin
bash prompt.

I've searched the lists and have seen several references to this problem
on NT, but found no clear solution.  Good People, do you have any
suggestions
on how to deal with this?

Dan

Re: [PORTS] Seeking advice about the dreaded stuck spinlock problem on NT

From
"Alexei Zakharov"
Date:
Do you run postgres with a patched ipc-daemon?  Unpatched daemon has a
semaphore bug which causes spinlocks.

Try to use this one (patched and compiled):
http://www.s34.co.jp/~luster/pgsql/cygwin32_ipc-1.03-bin-patched.tar.bz2

Alexei.

----- Original Message -----
From: "Dan Rosner" <rosner@scn.org>
To: <pgsql-ports@postgreSQL.org>
Sent: Wednesday, February 02, 2000 2:54 PM
Subject: [PORTS] Seeking advice about the dreaded stuck spinlock problem on
NT


> Does anyone have any advice/suggestions about how to deal with
> stuck spinlock problems while running postgres 6.5.3
> on cygwin B20?
>
> Background:
>
> Running on NT with sp6, I'm using active perl with DBI
> and DBD-PG to load data into postgres and update some tables
> every five minutes with a simple batch file that executes my
> perl scripts.  Everything seems to work well for a few hours,
> but then the backend dies with the dreaded...
>
> FATAL: s_lock(14490065) at spin.c:125, stuck spinlock. Aborting.
>
> After this happens, I'm unable to connect to the backend unless I
> kill and restart the ipc-daemon and postmaster again from the cygwin
> bash prompt.
>
> I've searched the lists and have seen several references to this problem
> on NT, but found no clear solution.  Good People, do you have any
> suggestions
> on how to deal with this?
>
> Dan
>
> ************