The patch I've applied 2 days ago doesn't work. And neither returning from inside the catch (it actually makes things even worse, but the problem is that I've not fully grasped all of Tigase internals yet)
Wojciech Kapcia wrote:
A few questions:
- does it happen every time?
It happens after some time, but it has always led me to restart Tigase.
- is the issue gone if you remove the additional field from the command?
Well, that might be a problem... the field is needed for authentication, so I can't do that on production where the users are "unstable" enough (e.g. I'm guessing this happens because of faulty mobile connections).
- have you made any other changes related to SessionManager handling and/or XMPPResourceConnection (I've briefly went over your github repository history but haven't noticed any significant changes)?
I'm writing some stress tests to run locally to see if I can reproduce the issue with "good" (see: local) connections. If I'll not be successful, I'll try something to simulate broken sockets: maybe I can achieve something with iptables rules, I don't know...