19 Sep
2003
19 Sep
'03
10:14 p.m.
Thanks, that worked. No memory leaks yet, but I have a bit bucket just in case.
-----Original Message-----
From: Andrei Pelinescu-Onciul [mailto:pelinescu-onciul@fokus.fraunhofer.de]
Sent: Friday, September 19, 2003 2:35 PM
To: Rork, Joseph (J.P.)
Cc: 'Serusers (serusers(a)lists.iptel.org)'
Subject: Re: [Serusers] No mem for sip_msg
On Sep 19, 2003 at 14:31, Rork, Joseph (J.P.) <jrork(a)ford.com> wrote:
What do you suggest it be increased too? Currently
the line reads
'#define PKG_MEM_POOL_SIZE 1024*1024'
Try 4096*1024, or 10240*1024 (10Mb).
As long as you have enough RAM you can increase it :-)
Andrei
-----Original Message-----
From: Andrei Pelinescu-Onciul
[mailto:pelinescu-onciul@fokus.fraunhofer.de]
Sent: Friday, September 19, 2003 2:18 PM
To: Rork, Joseph (J.P.)
Cc: 'Serusers (serusers(a)lists.iptel.org)'
Subject: Re: [Serusers] No mem for sip_msg
On Sep 19, 2003 at 14:06, Rork, Joseph (J.P.) <jrork(a)ford.com> wrote:
I've recently gone back to MySql v4.0, w/ SER
8.11. I'm trying to
log
in and I am seeing the following message in the syslog.
Sep 19 17:50:17 sip2 ser[2066]: ERROR: receive_msg: no mem for
sip_msg
I have plenty of memory in my system. Has anyone else run into this
problem, and if so, how did you fix it?
The short fix is edit config.h and change PKG_MEM_POOL_SIZE. Right now
is 1Mb, increase it.
Explanation:
ser uses 2 types of memory, shared and "local". The "local" memory
is
not shared between the processes and it's faster to allocate since we
don't need to lock on malloc. The shared mem. size can be increased
form the command line (-m size_in_mb). The default is 32Mb. The
"local" mem. maximum size is fixed to 1Mb (PKG_MEM_POOL_SIZE). We
haven't needed more so far. In your case probably ser uses a lot of
"local" mem. when it tries to load some of the databases in RAM on
init. This is a known problem, and in the future we'll solve it
without increasing PKG_MEM_POOL_SIZE, but for now this is the only
solution. (I assume this is the problem in your case, but you could
have also found a mem. leak, we would need the full logs and maybe a
debug build to say for sure).
Andrei