Hi again,
I missed the part where you said you increased the private memory to
100M :) .
The dr_reload command fetches one chunk at a time - you can modify the
number of rows fetched by setting fetch_rows parameter in dr_module .But
there could also be a leak if you say that this happens after a couple
of reloads.
I suggest you to do a monitoring of the private memory for the FIFO
process. You can use the following command:
/opensipsctl fifo get_statistics pkmem: | grep pkmem:1-/
and run this with a script and collect the data.
You can also use OpenSIPS-CP to monitor the pkgmem_real_used statistic.
Regards,
--
Anca Vamanu
OpenSIPS Developer
On 04/21/2011 04:56 PM, thrillerbee wrote:
I'm trying to track down the source of the following errors in syslog:
/usr/local/sbin/opensips[25490]: WARNING:core:fm_malloc: Not enough
free memory, will atempt defragmenation
/usr/local/sbin/opensips[25490]: ERROR:core:db_allocate_rows: no
memory left
/usr/local/sbin/opensips[25490]:
ERROR:db_postgres:db_postgres_convert_rows: no private memory left
What makes this even more unusual is that it will print these errors
at almost exactly 30-min intervals for several hours and then just
stop. For example, last night, it printed the above errors at:
23:56:00
00:26:08
00:56:08
01:26:03
01:56:05
02:26:03
02:56:01
03:26:02
and then it stopped...
I see no errors in postgresql logs and I haven't yet found any issues
with the actual performance of OpenSIPS. I have increased private
memory to 100MB and continue to see this randomly. This OpenSIPS
instance is using drouting module with about 16,000 lines in the
dr_rules table.
Any advice would be appreciated to help narrow down the source of
these messages.
Thanks,
Ryan
_______________________________________________
Users mailing list
Users@lists.opensips.org
http://lists.opensips.org/cgi-bin/mailman/listinfo/users
