2 Dec
2013
2 Dec
'13
3 p.m.
Dear Support,
I am having an "out of memory" problem while using kamailio with SEAS
module.
After using Kamailio with the SEAS module for few hours with a high load , I
am getting "out of memory" error that I believe is caused from a memory
leak.
I would like your help interpreting this error.
The output of kamailio -v is :
version: kamailio 4.0.4 (i386/linux) cabe58
flags: STATS: Off, USE_IPV6, USE_TCP, USE_TLS, TLS_HOOKS, USE_RAW_SOCKS,
DISABLE_NAGLE, USE_MCAST, DNS_IP_HACK, SHM_MEM, SHM_MMAP, PKG_MALLOC,
USE_FUTEX, FAST_LOCK-ADAPTIVE_WAIT, USE_DNS_CACHE, USE_DNS_FAILOVER,
USE_NAPTR, USE_DST_BLACKLIST, HAVE_RESOLV_RES
ADAPTIVE_WAIT_LOOPS=1024, MAX_RECV_BUFFER_SIZE 262144, MAX_LISTEN 16,
MAX_URI_SIZE 1024, BUF_SIZE 65535, DEFAULT PKG_SIZE 4MB
poll method support: poll, epoll_lt, epoll_et, sigio_rt, select.
id: cabe58
compiled on 14:23:27 Dec 2 2013 with gcc 4.1.2
kamctl ps :
Process:: ID=0 PID=31191 Type=attendant
Process:: ID=1 PID=31192 Type=udp receiver child=0 sock=193.100.200.18:5070
Process:: ID=2 PID=31193 Type=udp receiver child=1 sock=193.100.200.18:5070
Process:: ID=3 PID=31194 Type=udp receiver child=2 sock=193.100.200.18:5070
Process:: ID=4 PID=31195 Type=udp receiver child=3 sock=193.100.200.18:5070
Process:: ID=5 PID=31196 Type=slow timer
Process:: ID=6 PID=31197 Type=timer
Process:: ID=7 PID=31198 Type=ctl handler
Process:: ID=8 PID=31199 Type=MI FIFO
Process:: ID=9 PID=31200 Type=SEAS
Process:: ID=10 PID=31201 Type=tcp receiver (generic) child=0
Process:: ID=11 PID=31202 Type=tcp receiver (generic) child=1
Process:: ID=12 PID=31203 Type=tcp receiver (generic) child=2
Process:: ID=13 PID=31204 Type=tcp receiver (generic) child=3
Process:: ID=14 PID=31205 Type=tcp main process
kamcmd pkg.stats:
{
entry: 0
pid: 31191
rank: 0
used: 72480
free: 4091840
real_used: 102448
}
{
entry: 1
pid: 31192
rank: 1
used: 80624
free: 4083696
real_used: 110592
}
{
entry: 2
pid: 31193
rank: 2
used: 80624
free: 4083696
real_used: 110592
}
{
entry: 3
pid: 31194
rank: 3
used: 80624
free: 4083696
real_used: 110592
}
{
entry: 4
pid: 31195
rank: 4
used: 80624
free: 4083696
real_used: 110592
}
{
entry: 5
pid: 31196
rank: -1
used: 2276560
free: 4083744
real_used: 17770256
}
{
entry: 6
pid: 31197
rank: -1
used: 1787296
free: 4083744
real_used: 17280992
}
{
entry: 7
pid: 31198
rank: -2
used: 77760
free: 4086544
real_used: 107744
}
{
entry: 8
pid: 0
rank: 0
used: 89920
free: 4074352
real_used: 119936
}
{
entry: 9
pid: 0
rank: 0
used: 0
free: 0
real_used: 0
}
{
entry: 10
pid: 31201
rank: 5
used: 140240
free: 4023728
real_used: 170560
}
{
entry: 11
pid: 31202
rank: 6
used: 140240
free: 4023712
real_used: 170576
}
{
entry: 12
pid: 31203
rank: 7
used: 140240
free: 4023856
real_used: 170432
}
{
entry: 13
pid: 31204
rank: 8
used: 140240
free: 4023712
real_used: 170576
}
{
entry: 14
pid: 31205
rank: -4
used: 3877344
free: 4030176
real_used: 19269056
}
As you can see above , the output of the SEAS module is not showing (pid
31200)
Moreover , I can see that the real_used value in the kamcmd core.shmmem keep
on increasing , and free decrease until I finally get an error "out of
memory"
When my SEAS client exit the output of kamcmd core.shmmem :
{
total: 33554432
free: 16273440
used: 1787296
real_used: 17280992
max_used: 33553520
fragments: 8098
}
The traffic is a simple SIP MESSAGE from a seagull simulator, with a 200OK
reply from my SEAS client.
I appreciate your support,
Thank you in advance,
Ziad Habchi