Hi,
I have an installation of pretty stock, vanilla SEMS 1.6 (1.6:5a6542d)
doing a light, mostly transparent 'sbc' workload -- just signalling, no
RTP relay. The only non-transparent parts of the egress profile are:
---
# Egress SBC profile.
RURI=sip:[email protected]
From=<sip:$fU@$Ri:$Rp>
To=<sip:[email protected]>
Contact=<sip:$Ri>
Call-ID=$ci_bleg
append_headers="X-SRC-IP: $si\r\nX-SRC-Port: $sp\r\n"
next_hop_ip=xxx.xxx.xxx.xxx
next_hop_port=5060
next_hop_for_replies=yes
enable_rtp_relay=no
enable_auth=no
enable_session_timer=no
---
This installation handles a high volume of call setups, sometimes
peaking around a few hundred calls per second. The average is probably
around 50 CPS.
What I'm seeing is monotonically increasing heap size from SEMS. This
server has 16 GB of RAM. I restarted SEMS about 24 hours ago and its
resident memory size was 0.9%. Now it is at 7.9%:
root 2236 8.8 8.2 16975660 1341100 ? Sl Dec15 119:46
/usr/local/sbin/sems -P /var/run/sems.pid -P /var/run/sems.pid -f
/opt/evariste/sems/etc/sems.conf
It appears to be increasing slowly and not decreasing during low
activity periods.
SEMS was not being proactively monitored on this host; the problem was
detected because eventually, after a week or two of operation, it ran
out of memory and crashed.
Any suggestions on how to track this down or find the source of the leak
would be appreciated!
-- Alex
--
Alex Balashov | Principal | Evariste Systems LLC
303 Perimeter Center North, Suite 300
Atlanta, GA 30346
United States
Tel: +1-800-250-5920 (toll-free) / +1-678-954-0671 (direct)
Web: http://www.evaristesys.com/, http://www.csrpswitch.com/
_______________________________________________
Sems mailing list
[email protected]
http://lists.iptel.org/mailman/listinfo/sems
