[asterisk-bugs] [JIRA] (ASTERISK-27885) Calls to SIP phones fail to establish, debug log and tcpdump suggest a packet backlog

[Aragon Gouveia (JIRA)]

    [ https://issues.asterisk.org/jira/browse/ASTERISK-27885?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=243825#comment-243825 ] 

Aragon Gouveia commented on ASTERISK-27885:
-------------------------------------------

I've now enabled much more debugging and had it happen again. The problem seems to be due to a "sip reload" blocking chan_sip or at least having a much higher priority than other tasks. Asterisk takes about 10 seconds to process a SIP reload with our SIP config, and during this time a large backlog of unprocessed network traffic accumulates. When it finishes, all the backlog gets handled, including many SIP timeouts.

This wasn't a problem with 1.8. I can't see any evidence of DNS lookups being the cause, and our SIP config does not have any DNS names in it (every peer entry has host=dynamic).

I'm not sure if you want to consider this a bug or a regression, so I'll leave it up to you. From my side I'm just going to try eliminate or reduce the frequency of our SIP reloads, and for our next upgrade we'll migrate to chan_pjsip, which I'm hoping doesn't have the same problem.

Thank you!

> Calls to SIP phones fail to establish, debug log and tcpdump suggest a packet backlog
> -------------------------------------------------------------------------------------
>
>                 Key: ASTERISK-27885
>                 URL: https://issues.asterisk.org/jira/browse/ASTERISK-27885
>             Project: Asterisk
>          Issue Type: Bug
>      Security Level: None
>          Components: Channels/chan_sip/General
>    Affects Versions: 13.17.2, 13.20.0
>         Environment: Debian Stretch: Linux asterisk 4.9.0-4-amd64 #1 SMP Debian 4.9.65-3 (2017-12-03) x86_64 GNU/Linux
>            Reporter: Aragon Gouveia
>            Assignee: Aragon Gouveia
>            Severity: Minor
>              Labels: pjsip
>         Attachments: core-show-settings.txt, core-show-sysinfo.txt, debug.txt, sip-show-settings.txt, tcpdump.txt
>
>
> The PBX I operate receives calls from a telco via a DAHDI channel, and these typically all ring to SIP phones. At random times, a call will ring to a SIP phone, but answering the phone fails to establish the call. The ringing stops, but audio is just dead.
> I've been running Asterisk with debugging enabled on one SIP extension, while simultaneously running tcpdump in the background to capture the actual network communication. Every time it happens, tcpdump shows the phone's SIP responses reaching the Asterisk server, but the debug log shows the same packets being processed much later than the tcpdump timestamps. By the time Asterisk has processed the phone's response packets, the caller has already hungup or some other timeout has already occurred (depending on which packets get backlogged).
> Unfortunately I have no way to reliably reproduce this. It can take weeks and hundreds of successful calls to manifest. The PBX was recently upgraded from Debian Wheezy + Asterisk 1.8.10.0, and since then this problem started occurring (ie. since Asterisk 13 and Debian Stretch).
> Attaching files with more details:
> core-show-settings.txt - CLI output, "core show settings"
> core-show-sysinfo.txt - CLI output, "core show sysinfo"
> sip-show-settings.txt - CLI output, "sip show settings"
> debug.txt - Debug log, notice the silence between 08:30:07 and 08:30:15
> tcpdump.txt - Notice the delay in processing packets received between 08:30:10 and 08:30:17.



--
This message was sent by Atlassian JIRA
(v6.2#6252)