On Tue, 18 Jun 2019, Robert Dixon wrote:
This 30 second timeout problem has plagued Linphone users for years, and it
has been brought up on many lists, with no real solution.I was able to
partially solve it on a local network by eliminating the use of TCP in the
Network options. Just use
I was ecstatic. Then I went to use it on a larger network, and it failed
again in the same way. I do not understand this.
The issue is that the 200 ACK response is sent to the wrong ip.
Because linphone tells the peer (asterisk in this case) the wrong ip
in the SIP header. The SIP library does not trust what you tell it in
the config, and silently tries to second guess you when making a
connection. It will randomly choose some public IP on an
active interface to put in the SIP header for the peer to use. It
may or may not be actually routable...
In my case, I am using IPv6 on a private VPN network, and the SIP library
"knows" that I *must* be mistaken when I configure the private IP, and
chooses an arbitrary public IP instead. Even when this works, it
defeats the whole point of the VPN, and it usually doesn't work because the peer has no public IP6 - only the IP6 VPN.
For me, the work around is to disable IP6 on all public interfaces, leaving only the VPN interface during the call. (On linux, this
is "echo 1 >/proc/sys/net/ipv6/conf/ethx/disable_ipv6".) This
prevents the SIP library from finding any (bogus for this purpose)
public IPs to use instead of what I told it. This is *highly*
inconvenient (all public IP6 traffic is stopped during the call),
but a reliable work around.
The good part, is that I learned a little about SIP protocol tracing
I just had a thought. Are you using raw IPs? Maybe the SIP library
doesn't do this silly thing with DNS names.
Stuart D. Gathman <address@hidden
"Confutatis maledictis, flamis acribus addictis" - background song for
a Microsoft sponsored "Where do you want to go from here?" commercial._______________________________________________
Linphone-users mailing listaddress@hidden