[asterisk-bugs] [JIRA] (ASTERISK-27646) ICE fails with no candidate nominated

[Thomas Guebels (JIRA)]

    [ https://issues.asterisk.org/jira/browse/ASTERISK-27646?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=241948#comment-241948 ] 

Thomas Guebels commented on ASTERISK-27646:
-------------------------------------------

Reverting the changes made in ASTERISK-27088 makes the issue disappear for me. More precisely, since that patch has 2 parts, either setting the ICE role earlier as before or deactivating the other part about unidirectional ICE, fixes the issue.

I don't use the bundled version of pjproject. I can reproduce with an external pjproject 2.5.5 and even when upgrading to 2.7.1.

> ICE fails with no candidate nominated
> -------------------------------------
>
>                 Key: ASTERISK-27646
>                 URL: https://issues.asterisk.org/jira/browse/ASTERISK-27646
>             Project: Asterisk
>          Issue Type: Bug
>      Security Level: None
>          Components: Channels/chan_sip/General
>    Affects Versions: 13.18.0
>            Reporter: Thomas Guebels
>            Assignee: Thomas Guebels
>         Attachments: udp_C-0000003b.pcap
>
>
> Since the change made in ASTERISK-27088, we have intermittent ICE issues.
> The ICE negotiation fails on the asterisk side with all checks succeeded but none nominated.
> [2018-01-31 16:22:00.159361] DEBUG[4908][C-00000023] pjproject: 	icess0x7efe980 .ICE process complete, status=All ICE checklists failed (PJNATH_EICEFAILED)
> [2018-01-31 16:22:00.159377] DEBUG[4908][C-00000023] pjproject: 	icess0x7efe980 .Valid list
> [2018-01-31 16:22:00.159394] DEBUG[4908][C-00000023] pjproject: 	icess0x7efe980 . 0: [1] 100.64.72.5:26424-->172.16.39.121:51919 (not nominated, state=Succeeded)
> [2018-01-31 16:22:00.159414] DEBUG[4908][C-00000023] pjproject: 	icess0x7efe980 . 1: [1] 100.64.72.5:26424-->172.16.39.121:51919 (not nominated, state=Succeeded)
> On the browser (Chrome 64) side, it thinks that ICE succeeded but there is one-way voice and the call soon stops because asterisk doesn't reply to STUN pings anymore.
> It seems to be timing dependent as it can only be reproduced with at least 2 network paths with similar latency. When one is noticeably slower, it doesn't happen.



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