Hello, I’m encountering this weird behaviour when I retrieve a single endpoint using res_config_curl
as realtime backend. When Asterisk starts it performs a cURL request to fetch all endpoints and aors (POST /[endpoints|aor]/multi
), al these endpoints have the Not in use
state, this seems to be the desired behavior. But whenever a call is made to an non existing endpoint it performs a single cURL request (POST /[endpoints|aor]/single
), these endpoints show as unavailable until a reload of PJSIP is done. The endpoint however is being called, even with the Unavailable
state, but it does give problems when that endpoint is also a member of a queue.
I tried this exact scenario with an ODBC backend and this seems to be working as expecting where a single endpoint request does show up as Not in use
state.
Relevant configs:
modules.conf
:
[modules]
autoload=yes
preload=res_config_curl.so
sorcery.conf
:
[res_pjsip]
endpoint/cache=memory_cache,object_lifetime_stale=3600,full_backend_cache=yes
endpoint=config,pjsip.conf,criteria=type=endpoint
endpoint=realtime,ps_endpoints
aor/cache=memory_cache,object_lifetime_stale=3600,full_backend_cache=yes
aor=config,pjsip.conf,criteria=type=aor
aor=realtime,ps_aors
extconfig.conf
[settings]
ps_endpoints => curl,https://webserver/endpoints
ps_aors => curl,https://webserver/aors
queue_members => curl,https://webserver/queue_members
I’m not sure if this is a bug, if I’m doing something wrong or if this is the desired behaviour.