Hi Daniel,

Sorry for the delay in replying. I’ve attached blf.cap which shows the “light stays on” scenario. You’ll see that the final NOTIFY (packet 43) is a “confirmed” rather than a “terminated” as per the PUBLISH  (packet 41) that triggered it.

I’ve also attached presentity.txt which is the contents of the DB before the pcap was taken.

With regards to your question about the light going out after the entries in the table have expired, this does happen automatically. As soon as the table drops down to being empty (it takes a couple of minutes to fully clear), the light goes off. Subsequent calls will work correctly with BLF until it eventually stops working and the whole cycle repeats.

I have repeated the test with subs_db_mode 0 and the same results occur. This is, in fact, the state it was in when the attached pcap was taken.

Do you think the problem is in the cleanup of the data or in the way the active dialog is matched against the set of data in the table? Happy to prod through the code with gdb if you can point me in the direction of where to start looking.

Cheers

Phil

From: Daniel-Constantin Mierla [mailto:miconda@gmail.com]
Sent: 19 January 2016 23:26
To: Phil Lavin <phil.lavin@synety.com>; Kamailio (SER) - Users Mailing List <sr-users@lists.sip-router.org>
Subject: Re: [SR-Users] Strange PUA Behaviour

Can you get a pcap for a case with the new config? Being traveling, but maybe I get a chance to look at it soon.

Reading quickly on some docs, I noticed that subs_db_mode=3 and notifier_proceses=0 rise a conflict:

http://www.kamailio.org/docs/modules/devel/modules/presence.html#presence.p.notifier_processes

Can you try with subs_db_mode=0 and see if works different?

Cheers,
Daniel

On 19/01/16 20:35, Phil Lavin wrote:

Just to follow this up with more recent results. I’ve simplified things and have been testing calling from a Kamailio registered UA to a phone on the PSTN. There’s only 1 dialog in Kamailio for this and the results are just as strange. BLF works for a while and then breaks. In this particular case, the light does not go off after the call.

I have set subs_db_mode to 3 (new config below) and I can consistently reproduce the BLF light not turning off after the call has ended (the phone does not get sent a terminate). As soon as I truncate pua and presentity tables in the DB, the next call works fine.

modparam("presence", "subs_db_mode", 3)

modparam("presence", "notifier_processes", 0)

modparam("presence", "db_url", DBURL)

modparam("presence", "send_fast_notify", 0)

modparam("presence", "db_update_period", 20)

modparam("presence_xml", "db_url", DBURL)

modparam("presence_xml", "force_active", 1)

modparam("presence_dialoginfo", "force_single_dialog", 1)

modparam("presence_dialoginfo", "force_dummy_dialog", 1)

modparam("pua", "db_url", DBURL)

modparam("pua", "db_mode", 2)

modparam("pua", "update_period", 20)

modparam("pua_dialoginfo", "send_publish_flag", FLT_DLGINFO)

modparam("pua_dialoginfo", "override_lifetime", 124)

Before I truncate, the tables both a good number of rows in each (70ish).

Is it that they’re not being correctly cleaned up here?

Thanks

Phil

From: sr-users [mailto:sr-users-bounces@lists.sip-router.org] On Behalf Of Phil Lavin
Sent: 19 January 2016 18:11
To: miconda@gmail.com; Kamailio (SER) - Users Mailing List <sr-users@lists.sip-router.org>
Subject: Re: [SR-Users] Strange PUA Behaviour

Below is the relevant presence/pua stuff. Let me know if I should be examining other tables.

When the call ends, there are no dialogs remaining in the dialog table. A few things do hang around in the presentity and pua tables for a short period of time.

Regarding CLI changing, it does seem to do so. When the call is routed onto the billing platform, the CLI is changed to the local “extension” (e.g. 9989) on the leg that comes back to Kamailio, destined for the callee.

Regarding only advertising the leg going to the callee, not all calls will terminate on a UA registered against Kamailio (e.g. calls from a Kamailio registered UA to the PSTN). The more I think about it, determining the correct leg in all scenarios will be difficult.

-- 
Daniel-Constantin Mierla
http://twitter.com/#!/miconda - http://www.linkedin.com/in/miconda
Book: SIP Routing With Kamailio - http://www.asipto.com
http://miconda.eu