[asterisk-bugs] [JIRA] (ASTERISK-26875) MixMonitor randomly out of sync

[Aaron An (JIRA)]

     [ https://issues.asterisk.org/jira/browse/ASTERISK-26875?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Aaron An updated ASTERISK-26875:
--------------------------------

    Description: 
My asterisk use MixMonitor to record, and it is randomly out of sync，I try to look for the root cause and find the solution.
Both side are PJSIP channel.
The scenario is very simple:
Asterisk call Alias，Alias answer and then go to the DIALPLAN to start MixMonitor and Dial Bob，Bob response 183 with SDP but no RTP stream feed in time. the record is out of sync.

just look at the audiohook.c line 180. if other_factory_samples is 0 we never flush the factory although  "our_factory_ms - other_factory_ms"  may very large.

====================audiohook.c=============================
	our_factory_samples = ast_slinfactory_available(factory);
	our_factory_ms = ast_tvdiff_ms(*rwtime, previous_time) + (our_factory_samples / (audiohook->hook_internal_samp_rate / 1000));
	other_factory_samples = ast_slinfactory_available(other_factory);
	other_factory_ms = other_factory_samples / (audiohook->hook_internal_samp_rate / 1000);

	if (ast_test_flag(audiohook, AST_AUDIOHOOK_TRIGGER_SYNC) && other_factory_samples && (our_factory_ms - other_factory_ms > AST_AUDIOHOOK_SYNC_TOLERANCE)) {
		ast_debug(1, "Flushing audiohook %p so it remains in sync\n", audiohook);
		ast_slinfactory_flush(factory);
		ast_slinfactory_flush(other_factory);
	}

====================audiohook.c=============================

I think it should remove the condition "&& other_factory_samples". I have tested the scenario that out of sync recording, this fix it.

  was:
MixMonitor randomly out of sync when s
Both side are PJSIP channel.
Asterisk call Alias，Alias answer and then go to the DIALPLAN to start MixMonitor and Dial Bob，Bob response 183 with SDP but no RTP stream feed. the record is out of sync.

just look at the audiohook.c line 180. if other_factory_samples is 0 we never flush the factory although  "our_factory_ms - other_factory_ms"  may very large.

====================audiohook.c=============================
	our_factory_samples = ast_slinfactory_available(factory);
	our_factory_ms = ast_tvdiff_ms(*rwtime, previous_time) + (our_factory_samples / (audiohook->hook_internal_samp_rate / 1000));
	other_factory_samples = ast_slinfactory_available(other_factory);
	other_factory_ms = other_factory_samples / (audiohook->hook_internal_samp_rate / 1000);

	if (ast_test_flag(audiohook, AST_AUDIOHOOK_TRIGGER_SYNC) && other_factory_samples && (our_factory_ms - other_factory_ms > AST_AUDIOHOOK_SYNC_TOLERANCE)) {
		ast_debug(1, "Flushing audiohook %p so it remains in sync\n", audiohook);
		ast_slinfactory_flush(factory);
		ast_slinfactory_flush(other_factory);
	}

====================audiohook.c=============================

I think it should remove the condition "&& other_factory_samples". I have tested the scenario that out of sync recording, this fix it.


> MixMonitor randomly out of sync
> -------------------------------
>
>                 Key: ASTERISK-26875
>                 URL: https://issues.asterisk.org/jira/browse/ASTERISK-26875
>             Project: Asterisk
>          Issue Type: Bug
>      Security Level: None
>          Components: Applications/app_mixmonitor
>    Affects Versions: 13.8.0
>         Environment: CentOS6.5X64
>            Reporter: Aaron An
>
> My asterisk use MixMonitor to record, and it is randomly out of sync，I try to look for the root cause and find the solution.
> Both side are PJSIP channel.
> The scenario is very simple:
> Asterisk call Alias，Alias answer and then go to the DIALPLAN to start MixMonitor and Dial Bob，Bob response 183 with SDP but no RTP stream feed in time. the record is out of sync.
> just look at the audiohook.c line 180. if other_factory_samples is 0 we never flush the factory although  "our_factory_ms - other_factory_ms"  may very large.
> ====================audiohook.c=============================
> 	our_factory_samples = ast_slinfactory_available(factory);
> 	our_factory_ms = ast_tvdiff_ms(*rwtime, previous_time) + (our_factory_samples / (audiohook->hook_internal_samp_rate / 1000));
> 	other_factory_samples = ast_slinfactory_available(other_factory);
> 	other_factory_ms = other_factory_samples / (audiohook->hook_internal_samp_rate / 1000);
> 	if (ast_test_flag(audiohook, AST_AUDIOHOOK_TRIGGER_SYNC) && other_factory_samples && (our_factory_ms - other_factory_ms > AST_AUDIOHOOK_SYNC_TOLERANCE)) {
> 		ast_debug(1, "Flushing audiohook %p so it remains in sync\n", audiohook);
> 		ast_slinfactory_flush(factory);
> 		ast_slinfactory_flush(other_factory);
> 	}
> ====================audiohook.c=============================
> I think it should remove the condition "&& other_factory_samples". I have tested the scenario that out of sync recording, this fix it.



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