Hello Bogdan, An mar., déc 22, 2009, Bogdan-Andrei Iancu schrieb: >[email protected] wrote: >>>> But this is maybe a clue. It would seem that something in TLS >>>> writing has changed between these two versions, maybe fundementally? >>>> >>> 1.3 was doing infinite loop (for write and read), leading sometime to >>> blocking. >>> >> That was a painful part of 1.3, so good that the counter is there >> now. I guess you're saying that the same TLS problems existed in >> 1.3 as well, but they were masked by retries (maybe thousands.) >> >yes, that is correct. > Can it be that when the internal OpenSIPS TCP lifetime counter is set to the registration interval using the tcp_persistent_flag that this counter is used even when the registration forcefully expires?
What I mean by forcefully is: Some IP phones don't wait for a registration period to time out. Instead they wait for 1/2 of the expiry period and then send a new REGISTER with a header 'Expires: 0' to force the registration to timeout. Then they immediately send a new REGISTER with a normal expiry value to obtain a new registration. ...so if an expiry time is 10 minutes, after only 5 minutes the UAC invalidates the registration and makes a new one. I'm wondering if OpenSIPS tries opening TCP connections using the value of 10 minutes with a UAC which is no longer in the location table because the AOR was removed (due to the principle described above.) Possible? Brian _______________________________________________ Users mailing list [email protected] http://lists.opensips.org/cgi-bin/mailman/listinfo/users
