Fax API and duplicate UUID

[Arek Fryz]

Hi all!

I'm wondering if anyone who use fax api to send bulk faxes see: 

[CRIT] switch_core_session.c:2317 Duplicate UUID!
[CRIT] mod_sofia.c:4232 Error Creating Session

I see that on my FS and looks like it tries to send the same fax job again and again. First re-try on job being in processing state happens couple seconds after first job started. I checked database and I see status: 'processing' in there.

Regards,

Arek Fryz

[Luis Azedo]

Hi,

this may happen if you bounce back the fax call.

--
You received this message because you are subscribed to the Google Groups "2600hz-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email to 2600hz-dev+...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Arek Fryz]

I'm not bouncing back,  all faxes go offnet. Also I noticed it really tries to duplicate tasks and if I don't get duplicate uuid it rings on number being processed at the moment and gets busy signal.. It does not happen in all jobs but on most of them.

Arek

Sent from REMAC Samsung Galaxy cell phone.

[Luis Azedo]

going offnet doesn't mean it will not bounce back. if you send a fax to a number in the system it will bounce back in some way, right ? 

i have only seen this when i was simulating load on the fax whapp and was trying to send faxes between accounts without having a second freeswitch. calls were sent from freeswitch to kamailio and then back to freeswitch leading to the duplicate uuid error. i setup a vanilla freeswitch not connected to kazoo to actually do the load simulation.

if that is not the case, is this a all-in-one setup ? a cluster ?

 

[Arek Fryz]

Luis,

I'm sending faxes to numbers that are not on any other account. It looks like kazoo does not recognize already ongoing process and try to start it again or try it again. I tried to disable all FS server and work with one - same thing. I tried to disable all fax app on all whapps but one (to make sure jobs are not being sent from different whapps servers at the same time), but this did not help too. Also, there is retry_after that looks like is being ignored. I played with job_timeout and wait_for_fax_timeout_ms but that did not help.

There is one more thing I will have to test.. I'm forcing specific fs_path on outbound, maybe that cause issues. I will play with it tonight.

Regards,

Arek Fryz

[Luis Azedo]

what version are you on ? 3.16, 3.18, master ? any changes you may have applied ?

fax has been pretty stable now.

fs_path on outbound ? what do you mean ?

[Arek Fryz]

I'm on 3.16 - no custom changes to this installation. fs_path to force SIP signaling go via specific kamailio server

Regards,

Arek Fryz

[Luis Azedo]

there must be something you're doing thats causing this.

don't know how can i help you with this.

[Luis Azedo]

btw,

what changes did you make to force fs_path on the originate command ? and.., is the kamailio part of the cluster? is the freeswitch part of dispatch_list in kamailio ? can you share your topology?( change public ip addresses)

since you're running a customized version of 3.16, i'm not sure how to help you.

did you try to run it without your changes ?

you can attach you freeswitch debug log from the beginning of the originate command (maybe 1 min earlier then that) until the error occurs.

[Luis Azedo]

kamailio log would also help.

cheers

[Arek Fryz]

Finally got some time to analyze logs and I found out this:

failed to execute api command originate on freeswitch@[...] exit {timeout,{gen_server,call,[{mod_kazoo,[..]

originate command actually happened and call was connected but for some reason whapps did not see it and sent another originate command to FS and that resulted in duplicated UUID. 

In error body I see that timeout was set to 10 seconds. Can I adjust timeout? 

[Arek Fryz]

I made quick patch that fixes duplicated UUID in my case. With this patch I'm able to control timeout on originate cmd and that will also control for how long we wait till called party answers our call. Useful with fax machines that are set to autonawer after long number of rings. In my case it was set to 10 seconds by ecallmgr and I think that included connection time + ringing time.

To control timeout you need to add "endpoint_timeout": XX into system_config/fax where XX is number of seconds. (and flush whapps_config)

I will appreciate review and suggestions. 

Here is links to commit: https://github.com/remac/kazoo/commit/b359222b984c59811a78de955abf8b7f1a90180d

Regards,

Arek Fryz