[asterisk-bugs] [JIRA] (ASTERISK-26842) Websocket becomes disconnected when trying to place call from browser

Friendly Automation (JIRA) noreply at issues.asterisk.org
Wed Mar 8 10:08:10 CST 2017


    [ https://issues.asterisk.org/jira/browse/ASTERISK-26842?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=235620#comment-235620 ] 

Friendly Automation commented on ASTERISK-26842:
------------------------------------------------

Change 5137 merged by zuul:
res_http_websocket: Fix faulty read logic.

[https://gerrit.asterisk.org/5137|https://gerrit.asterisk.org/5137]

> Websocket becomes disconnected when trying to place call from browser
> ---------------------------------------------------------------------
>
>                 Key: ASTERISK-26842
>                 URL: https://issues.asterisk.org/jira/browse/ASTERISK-26842
>             Project: Asterisk
>          Issue Type: Bug
>      Security Level: None
>          Components: Resources/res_http_websocket
>    Affects Versions: GIT
>            Reporter: Mark Michelson
>
> When using a WebRTC SIP client (Firefox browser, SIPml5 demo), I noticed that I could register, but when I would attempt to call, I would see
> {noformat}
> [Mar  7 19:24:26] WARNING[3005]: res_http_websocket.c:526 ws_safe_read: Websocket seems unresponsive, disconnecting ...
> [Mar  7 19:24:26] ERROR[3005]: tcptls.c:744 ast_tcptls_close_session_file: ast_tcptls_close_session_file invoked on session instance without file or file descriptor
> {noformat}
> After investigating further, I found that the issue was centered around a coding error introduced to res_http_websocket.c when the iostreams change was merged. The problem was that on an EINTR read from the websocket, the expected length was being adjusted as if a partial read had occurred. By adding a simple if block around the expected length adjustment, the problem is bypassed and websockets work fine.



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



More information about the asterisk-bugs mailing list