I have three asterisk machines and three SBCs. The cluster part is working perfectly fine but if one SBC is down, the dialplan keeps stuck for a long time on Dial. After some time, the Dial fails as CONGESTION and fallthrough takes the next one.
How can I set the time to wait for an initial reply to some seconds? The SBC has an interface directly on the same network and will reply immediately if it’s health is good.
There is a timeout option on Dial but that is used for ring timeout.
Are you getting no reply to critical request messages? (I’m not sure how PJSIP presents these.)
Correct, INVITE is going to be sent but ARP on the network is unable to locate the node (as it is down):
Everyone is busy/congested at this time (1:0/0/1)
That timer is normally adaptive, based on qualify, but, if you used qualify, it would know the peer was down, and wouldn’t even need to send anything.
These are the timers used when there is no adaptation and no qualify:
; SYNOPSIS: Options that apply to the SIP stack as well as other system-wide settings
;timer_t1=500 ; Set transaction timer T1 value milliseconds (default: "500")
;timer_b=32000 ; Set transaction timer B value milliseconds (default: "32000")
Definitions can be found in the SIP RFC.
Qualify sounds good, I did the same on freeswitch machines.
I will try to get this working with IP auth.
The chan_pjsip module does not currently dynamically alter the value based on the result of qualify.