Debugging BLF Problems
Revision as of 22:59, 25 February 2018 by Dcunningham (talk | contribs)
To debug BLF we need the following traces.
They are for the time period where the monitored phone (phone A) receives a call from another phone (phone B), but the subscribing phone (phone C)'s BLF light is not set correctly:
- A ngrep trace on the Kamailio/OpenSIPS server showing all traffic to and from phones A and C. It might be easiest to take a PCAP of all network traffic passing through the server using:
ngrep -d any -O /tmp/ticket-XXXX.pcap > /dev/null
- An Asterisk SIP and AGI trace of the call from phone B to phone A. You can take this using the Reports -> Traces menu at the system level in Enswitch.
- A Mysql general log showing all SQL queries. See: https://dev.mysql.com/doc/refman/5.7/en/query-log.html
Please also let us know:
- The number of phone A (monitored phone).
- The number of phone B (calling phone).
- The number of phone C (subscribing phone).
- The exact start time of the call which triggered the BLF activity.
- A description of what exactly was not correct. For example: "When phone A started to ring, the BLF light on phone C did not light up, and still did not light when phone A picked up the call".
It helps a lot fix the issue if the call uses the simplest possible scenario, so for example:
- Have just one phone (C) with a BLF subscription for the monitored phone A, and have no BLF subscriptions on phone A or B. This way a NOTIFY for phone C from phone A can't be confused with a NOTIFY for phone A from phone C.
- Use a direct phone-to-phone call, not hunt groups or queues (unless the problem only happens using hunt groups or queues).
- Take the traces when the system is as quiet as possible, for example late in the evening.
- Include just one call which reproduces the problem in the traces. Usually only one example is necessary, but if taking multiple calls put them in separate trace files.