Bogdan,
I upgraded to the current cvs version, and I think that has fixed it. With no change to openser.cfg, it's been running for about 3 hours now without any reported memory issues.
Douglas.
-----Original Message----- From: Bogdan-Andrei Iancu [mailto:bogdan@voice-system.ro] Sent: Monday, June 26, 2006 10:20 AM To: Douglas Garstang Cc: Users@openser.org Subject: Re: [Users] Out of Memory
Ok Douglas - please let me know if the latest changes in resolver have any effect on the leak you observed.
regards, bogdan
Douglas Garstang wrote:
Bogdan,
Don't appear to have ever received that message.
To answer your questions: We are not using ENUM at all. We aren't using our own module
for DNS lookups.
I will try the latest cvs.
Doug
-----Original Message----- From: Bogdan-Andrei Iancu [mailto:bogdan@voice-system.ro] Sent: Monday, June 26, 2006 8:33 AM To: Douglas Garstang Cc: Users@openser.org Subject: Re: [Users] Out of Memory
See the list. I sent you a reply : http://www.openser.org/pipermail/users/2006-June/005321.html
regards, bogdan
Douglas Garstang wrote:
-----Original Message----- From: Douglas Garstang Sent: Thursday, June 22, 2006 12:09 PM To: Bogdan-Andrei Iancu Cc: Users@openser.org Subject: RE: [Users] Out of Memory
-----Original Message----- From: Douglas Garstang Sent: Wednesday, June 21, 2006 12:48 PM To: Bogdan-Andrei Iancu Cc: Users@openser.org Subject: RE: [Users] Out of Memory
Bogdan,
I finally managed to upload the memory dumps to pastebin.
Links are:
Hopefully you can access them.
Memory dump right after OpenSER was started: http://pastebin.com/723872 Memory dump after OpenSER running for 20min: http://pastebin.com/723890 Memory dump after OpenSER running for 50min: http://pastebin.com/723902
The problem started to occur between the 20 and 50 minute samples. About 20 calls had been processed in that time.
So... I'm wondering if anything has been determined with this...
Hi Bogdan. Did you manage to find anything with this?
Douglas.
Douglas,
please keep an eye on it. If actually the fix solved the problem, it means you have a really strange DNS system (it sends in response non SRV records to SRV requests)...
so, please let it running and inspect the mem. logs from time to time to see if any info.
thanks and regards, bogdan
Douglas Garstang wrote:
Bogdan,
I upgraded to the current cvs version, and I think that has fixed it. With no change to openser.cfg, it's been running for about 3 hours now without any reported memory issues.
Douglas.
-----Original Message----- From: Bogdan-Andrei Iancu [mailto:bogdan@voice-system.ro] Sent: Monday, June 26, 2006 10:20 AM To: Douglas Garstang Cc: Users@openser.org Subject: Re: [Users] Out of Memory
Ok Douglas - please let me know if the latest changes in resolver have any effect on the leak you observed.
regards, bogdan
Douglas Garstang wrote:
Bogdan,
Don't appear to have ever received that message.
To answer your questions: We are not using ENUM at all. We aren't using our own module
for DNS lookups.
I will try the latest cvs.
Doug
-----Original Message----- From: Bogdan-Andrei Iancu [mailto:bogdan@voice-system.ro] Sent: Monday, June 26, 2006 8:33 AM To: Douglas Garstang Cc: Users@openser.org Subject: Re: [Users] Out of Memory
See the list. I sent you a reply : http://www.openser.org/pipermail/users/2006-June/005321.html
regards, bogdan
Douglas Garstang wrote:
-----Original Message----- From: Douglas Garstang Sent: Thursday, June 22, 2006 12:09 PM To: Bogdan-Andrei Iancu Cc: Users@openser.org Subject: RE: [Users] Out of Memory
>-----Original Message----- >From: Douglas Garstang >Sent: Wednesday, June 21, 2006 12:48 PM >To: Bogdan-Andrei Iancu >Cc: Users@openser.org >Subject: RE: [Users] Out of Memory > > >Bogdan, > >I finally managed to upload the memory dumps to pastebin. > > > >
Links are:
>Hopefully you can access them. > >Memory dump right after OpenSER was started: >http://pastebin.com/723872 >Memory dump after OpenSER running for 20min: >http://pastebin.com/723890 >Memory dump after OpenSER running for 50min: >http://pastebin.com/723902 > >The problem started to occur between the 20 and 50 minute >samples. About 20 calls had been processed in that time. > > > > > > So... I'm wondering if anything has been determined with this...
Hi Bogdan. Did you manage to find anything with this?
Douglas.
Hi Bogdan.
I have the same problem as Douglas, but to apply resolve.c patch don't solve the problem. I'm using version 1.1.0.
Memory debug info is available at http://vorigon.com/mhack/openser/ the problem occur only in threads 23208, 23209, 23210 and 23211.
I stopped the load during 1 hour but when I restarted the problem came back to happen in the same threads.
My configuration file is in the same location.
Regards,
Marcos Hack.
Sorry, obviously the problem occur only in threads of active interface.
Marcos Hack wrote:
Hi Bogdan.
Memory debug info is available at http://vorigon.com/mhack/openser/ the problem occur only in threads 23208, 23209, 23210 and 23211.
I stopped the load during 1 hour but when I restarted the problem came back to happen in the same threads.
After a lot of tests I have discovered that the problem is in nathelper module.
When using this module the process alloc all available pkg memory calling function build_req_buf_from_sip_req() from msg_translator.c as you can see in debug info at [1].
Is this a memory leak in nathelper module?
Somebody can help me with this problem?
[1] http://vorigon.com/mhack/openser/20060720-1744/mem_dbg.tar.gz
Marcos Hack wrote:
Hi Bogdan.
I have the same problem as Douglas, but to apply resolve.c patch don't solve the problem. I'm using version 1.1.0.
Memory debug info is available at http://vorigon.com/mhack/openser/ the problem occur only in threads 23208, 23209, 23210 and 23211.
I stopped the load during 1 hour but when I restarted the problem came back to happen in the same threads.
My configuration file is in the same location.
Hi Marcos,
thanks for report. are you sure it is about the nathelper module? I ask because this module does not use the build_req_buf_from_sip_req() function. A fast search shows that the function is used only from two places: forward.c:322 modules/tm/t_fwd.c:173
It will be very helpful if you could post the entire log - from startup - and not only the mem dumps; I can look where the chunks are allocated.
regards, bogdan
Marcos Hack wrote:
After a lot of tests I have discovered that the problem is in nathelper module.
When using this module the process alloc all available pkg memory calling function build_req_buf_from_sip_req() from msg_translator.c as you can see in debug info at [1].
Is this a memory leak in nathelper module?
Somebody can help me with this problem?
[1] http://vorigon.com/mhack/openser/20060720-1744/mem_dbg.tar.gz
Marcos Hack wrote:
Hi Bogdan.
I have the same problem as Douglas, but to apply resolve.c patch don't solve the problem. I'm using version 1.1.0.
Memory debug info is available at http://vorigon.com/mhack/openser/ the problem occur only in threads 23208, 23209, 23210 and 23211.
I stopped the load during 1 hour but when I restarted the problem came back to happen in the same threads.
My configuration file is in the same location.
Devel mailing list Devel@openser.org http://openser.org/cgi-bin/mailman/listinfo/devel
-
Bogdan-Andrei Iancu -
Douglas Garstang -
Marcos Hack