Problem with webrtc

Hi,
I have an issue with webrtc clients. All is working very well when I have logged about 20 client, when this number going up asterisk stop respond on any action. I tried increase session_limit to 1000 in http.conf but no result.

Its start going wrong in 20-30 minutes after Asterisk start

Can you define “stop respond on any action”?

Does the CLI stop accepting commands? Do calls fail?

Hi,
Its mean I can NOT make any register/call etc. CLI works ok

And which channel driver are you using?

i am using only pjsip.
pjsip are registered to voip providers and clients via wss to asterisk

You would need to collect a backtrace[1] to see if Asterisk is actually deadlocked. What version of Asterisk is in use?

[1] https://wiki.asterisk.org/wiki/display/AST/Getting+a+Backtrace#GettingaBacktrace-Runningast_coredumperfordeadlocks,taskprocessorbackups,etc.

As I understand I should do this when situation will exist, right?
I am using

serv250018*CLI> core show version
Asterisk 17.8.1 built by root @ serv250018 on a x86_64 running Linux on 2020-11-07 16:42:30 UTC

It can only be done when the issue exists, as well if there is an issue then Asterisk 17 will not receive the fix.

So, Only 18 will receive fix ?

Current supported Asterisk versions are on the wiki[1]. 16 and 18 currently receive bug fixes.

[1] https://wiki.asterisk.org/wiki/display/AST/Asterisk+Versions

i upgraded Asterisk to 18.0.1, still this same issue

dmesg shows this

[Tue Nov 17 14:56:01 2020] conntrack: generic helper won’t handle protocol 47. Please consider loading the specific helper module.
[Tue Nov 17 17:32:05 2020] asterisk[5829]: segfault at 18 ip 00007f6743ab18b8 sp 00007f6693492ba0 error 4 in libasteriskpj.so.2[7f67439d7000+ea000]
[Tue Nov 17 17:32:05 2020] Code: 89 45 40 c7 45 48 01 00 00 00 eb d0 66 0f 1f 44 00 00 41 54 4c 8d 67 50 55 53 48 89 fb 48 8b 6f 58 49 39 ec 74 16 0f 1f 40 00 <48> 8b 7d 18 e8 cf 6b f2 ff 48 8b 6d 08 49 39 ec 75 ee 48 83 7b 40

That would be caused by a crash somewhere. You would need to follow the instructions I previously posted for getting a backtrace in the case of a deadlock or crash.