Hi,
I have been testing 'DNS SRV', backup server and t_replicate with two 'ser'servers, BT101(firmware:1.0.7.11), and two GXP2000s(1.0.1.9).
The t_replicate works fine between sip1(main) and sip2(backup) servers.
Then, I stopped 'ser' on sip1 server, three UAs sent REGISTERs not to sip1 but to sip2. I tested make calls between three UAs, sip2 took care of sip calls, worked fine.
After that, I restarted 'ser' on sip1. But the three UAs still send REGISTER to sip2, eventhough sip1 came back and became alive.
I set up 5 minutes intervals for these theree UAs for REGISTER.
So, I stopped 'ser' on sip2 for 10-15 minutes, three UAs sent REGISTERs to sip1 again and tryed to 't_replicate' to sip2, these were failed as 'destination unreachable'.
Then I restarted 'ser' on sip2 again. After REGISTERed on sip1, t_replicate send REGISTERs from sip1 to sip2. Now, everything works fine.
Is this server side opration normal? Or these UAs are still bad boys?
Regards,
Zen
Hi,
http://www.voip-info.org/wiki/view/SRV+implementations says bad clients:
- Grandstream GXP-2000, 1.0.1.12 uses first SRV, no failover, ignores ICMP
- Grandstream BR-102, 1.0.6.7 uses first SRV, no failover, ignores ICMP
GXP2000(1.0.1.9) and BT101(1.0.7.11) still use first SRV,no failover.
So workaround is(if UAs are set 5min REGISTER intervals):
stop ser1 stop ser2 | | V V ....|<-- 5 min-->|<--sip2-->|<-- 5min -->|<--sip1--.....-> |service stop|service |service stop|service A |continue | B |continue
During service stop time A and B, users can not make calls.
Is there another way to reduce service time time?
Regards,
Zen
-
Zen Kato