I am trying to implement a smart DNS failover for HA freeswitch, and it seems that the Jitsi client fails to notice DNS changes as long as it is connected. If it is connected to Host A and the freeswitch service on Host A stops, Jitsi desktop client will properly reregister to the new address at Host B. If however it is connected to Host A, and DNS changes to Host B (ie: due to web service down but freeswitch still live, or for reverting to primary after failure event is over) Jitsi will stay connected to Host B until explicitly disconnected and reconnected client-side.
Is this a bug? Is there a way to force Jitsi desktop to occasionally reregister (AND check DNS againat that time, respecting TTLs)? I have tried selecting the REGISTER keepalive method in hopes that could help, but alas it does not.
Currently when a Protocol Provider is registering it gets the nextDNSRecord and connects to it and this is what is used through the connected stated of the provider, if it fails or goes off and on again a new DNS query will be done. So what you observe is how things work at the moment.
It can be changed, I suppose, when re-registering to always make DNS query … so any PRs are welcome.