I wonder if anybody is using PJSIP realtime in production environment?
I’ve started to play with it and encountered many problems. Here’s my config:
ps_endpoints => pgsql,users,pjsip_endpoints_v
pjsip_endpoints_v is postgresql view.
1. The biggest problem: if I have small number of endpoints (roughly up to a 100) then asterisk loads ok and pjsip seems to be working ok (with other problems described below). If I have larger number of endpoints (several hundred) then intermittently (but often) asterisk just hangs during loading. Attempting to start asterisk with console (-c) it never reaches the user prompt. pjsip isn’t functional (doesn’t reply to any sip messages). the only way out is to kill asterisk. It looks like I’m hitting some limit here. Am I doing something wrong? Is there any config option I’m missing?
2. When it loads ok then it performs initial load of all endpoints individually. Looking at postgres log it does the following:
SELECT * FROM pjsip_endpoints_v WHERE id LIKE ‘%’ ORDER BY id SELECT * FROM pjsip_endpoints_v WHERE id = ‘ep1’
SELECT * FROM pjsip_endpoints_v WHERE id = ‘ep2’
… SELECT * FROM pjsip_endpoints_v WHERE id = ‘epN’
Needless to say it’s extremely inefficient with larger number of endpoints. After this initial load it seems to work correctly – loading endpoints on demand and caching them if I configure caching. Is there a way to disable this initial load? I want it to load endpoints on demand only.
3. When pjsip receives sip message and tries to match it to endpoint by ‘From’ username it initially performs lookup for ‘username@domain’ and if it fails it falls back to lookup by username only. Looking at the postgres log it looks like the following:
SELECT * FROM pjsip_endpoints_v WHERE id = ‘ep1@domain’
SELECT * FROM pjsip_endpoints_v WHERE id = ‘ep1’
In my environment the domain part may be different (depending on which proxy the user is terminated) and I want to perform lookup on userpart only. Is there a way to tell pjsip to ignore the domain? Otherwise it doubles the number of queries per request and that’s again extremely inefficient.
I’d appreciate any help and/or pointers. Thanks a lot,