Dear,
I'm using siptrace within a python handler in the KEMI framework. KSR.siptrace.sip_trace_dst("sip:<myhomerip>:9060")
I notice an ever increasing usage of memory, until the system runs out of memory. (definition of a memory leak :) ). Looking at the memory information kamailio is giving me I notice this:
May 14 05:49:48 server kamailio[36452]: NOTICE: PYT {INVITE} {1769595029-1513210001-1955500877} {1} : qm_sums: count= 79286 size= 5812048 bytes from siptrace: siptrace.c: ki_sip_trace_dst_cid(729) May 14 05:51:17 server kamailio[36452]: NOTICE: PYT {REGISTER} { 4_1419955802@192.168.0.153} {2582} : qm_sums: count= 79335 size= 5814840 bytes from siptrace: siptrace.c: ki_sip_trace_dst_cid(729)
I notice this on kamailio 5.1.8 (from a debian repo), and on kamailio 5.2.2 (from source). I can reproduce the problem very easily, and I can apply patches in the source to try out things.
Could I be wrongly using siptrace? Can I give you more debugging info?
Kind regards, Davy
Hi Davy,
we're using siptrace with kemi and python 2. I can not confirm your problem on kamailio 5.2.2.
In difference to your script, however, i'm not setting a target address in the invocation. Instead the cfg script sets modparam("siptrace", "duplicate_uri", "sip:<myhomerip>:9060") and uses KSR.siptrace.sip_trace()
Maybe it makes a difference, can you try it out?
Regards,
Thomas
----- Ursprüngliche Mail -----
Von: "davy van de moere" davy.van.de.moere@gmail.com An: "Kamailio (SER) - Users Mailing List" sr-users@lists.kamailio.org Gesendet: Dienstag, 14. Mai 2019 07:57:05 Betreff: [SR-Users] Possible memory leak in siptrace module in kamailio 5.1/5.2
Dear,
I'm using siptrace within a python handler in the KEMI framework. KSR.siptrace.sip_trace_dst("sip:<myhomerip>:9060")
I notice an ever increasing usage of memory, until the system runs out of memory. (definition of a memory leak :) ). Looking at the memory information kamailio is giving me I notice this:
May 14 05:49:48 server kamailio[36452]: NOTICE: PYT {INVITE} {1769595029-1513210001-1955500877} {1} : qm_sums: count= 79286 size= 5812048 bytes from siptrace: siptrace.c: ki_sip_trace_dst_cid(729) May 14 05:51:17 server kamailio[36452]: NOTICE: PYT {REGISTER} { [ mailto:4_1419955802@192.168.0.153 | 4_1419955802@192.168.0.153 ] } {2582} : qm_sums: count= 79335 size= 5814840 bytes from siptrace: siptrace.c: ki_sip_trace_dst_cid(729)
I notice this on kamailio 5.1.8 (from a debian repo), and on kamailio 5.2.2 (from source). I can reproduce the problem very easily, and I can apply patches in the source to try out things.
Could I be wrongly using siptrace? Can I give you more debugging info?
Kind regards, Davy
Kamailio (SER) - Users Mailing List sr-users@lists.kamailio.org https://lists.kamailio.org/cgi-bin/mailman/listinfo/sr-users
Actually so far it seems using KSR.siptrace.sip_trace("sip:<myhomerip>:9060") instead of KSR.siptrace.sip_trace_dst("sip:<myhomerip>:9060"), fixes it ! I'm rolling it out on more systems to be sure, and wait a bit longer to be very sure. But thx for the help!
Op di 14 mei 2019 om 09:28 schreef Thomas Weber thomas.weber@pascom.net:
Hi Davy,
we're using siptrace with kemi and python 2. I can not confirm your problem on kamailio 5.2.2.
In difference to your script, however, i'm not setting a target address in the invocation. Instead the cfg script sets modparam("siptrace", "duplicate_uri", "sip:<myhomerip>:9060") and uses KSR.siptrace.sip_trace()
Maybe it makes a difference, can you try it out?
Regards,
Thomas
----- Ursprüngliche Mail -----
Von: "davy van de moere" davy.van.de.moere@gmail.com An: "Kamailio (SER) - Users Mailing List" sr-users@lists.kamailio.org Gesendet: Dienstag, 14. Mai 2019 07:57:05 Betreff: [SR-Users] Possible memory leak in siptrace module in kamailio
5.1/5.2Dear,
I'm using siptrace within a python handler in the KEMI framework. KSR.siptrace.sip_trace_dst("sip:<myhomerip>:9060")
I notice an ever increasing usage of memory, until the system runs out of memory. (definition of a memory leak :) ). Looking at the memory
information
kamailio is giving me I notice this:
May 14 05:49:48 server kamailio[36452]: NOTICE: PYT {INVITE} {1769595029-1513210001-1955500877} {1} : qm_sums: count= 79286 size=
5812048
bytes from siptrace: siptrace.c: ki_sip_trace_dst_cid(729) May 14 05:51:17 server kamailio[36452]: NOTICE: PYT {REGISTER} { [ mailto:4_1419955802@192.168.0.153 | 4_1419955802@192.168.0.153 ] }
{2582} :
qm_sums: count= 79335 size= 5814840 bytes from siptrace: siptrace.c: ki_sip_trace_dst_cid(729)
I notice this on kamailio 5.1.8 (from a debian repo), and on kamailio
5.2.2
(from source). I can reproduce the problem very easily, and I can apply
patches
in the source to try out things.
Could I be wrongly using siptrace? Can I give you more debugging info?
Kind regards, Davy
Kamailio (SER) - Users Mailing List sr-users@lists.kamailio.org https://lists.kamailio.org/cgi-bin/mailman/listinfo/sr-users
Kamailio (SER) - Users Mailing List sr-users@lists.kamailio.org https://lists.kamailio.org/cgi-bin/mailman/listinfo/sr-users
Great that you can use this workaround. Although i'm a bit confused because KSR.siptrace.sip_trace() is not supposed to accept a string parameter.
Can you file a bug on github regarding sip_trace_dst()?
----- Ursprüngliche Mail -----
Actually so far it seems using KSR.siptrace.sip_trace("sip:<myhomerip>:9060") instead of KSR.siptrace.sip_trace_dst("sip:<myhomerip>:9060"), fixes it ! I'm rolling it out on more systems to be sure, and wait a bit longer to be very sure. But thx for the help!
Kamailio (SER) - Users Mailing List sr-users@lists.kamailio.org https://lists.kamailio.org/cgi-bin/mailman/listinfo/sr-users
Hello,
On 15.05.19 09:43, Thomas Weber wrote:
Great that you can use this workaround. Although i'm a bit confused because KSR.siptrace.sip_trace() is not supposed to accept a string parameter.
If the parameter is given, then it is ignored -- with scripting languages allowing dynamic number of parameters is not easy to detect at init/startup if a function has too many params.
Anyhow, in this case I discovered that a local variable was not freed - I pushed a patch to branch 5.2, master branch was changed a lot and doesn't have the same kind of code. Can anyone test and report if works ok or not?
Cheers, Daniel
Can you file a bug on github regarding sip_trace_dst()?
----- Ursprüngliche Mail -----
Actually so far it seems using KSR.siptrace.sip_trace("sip:<myhomerip>:9060") instead of KSR.siptrace.sip_trace_dst("sip:<myhomerip>:9060"), fixes it ! I'm rolling it out on more systems to be sure, and wait a bit longer to be very sure. But thx for the help!
Kamailio (SER) - Users Mailing List sr-users@lists.kamailio.org https://lists.kamailio.org/cgi-bin/mailman/listinfo/sr-users
Kamailio (SER) - Users Mailing List sr-users@lists.kamailio.org https://lists.kamailio.org/cgi-bin/mailman/listinfo/sr-users
-
Daniel-Constantin Mierla -
davy van de moere -
Thomas Weber