That happened on my server last week. I had to restart the server and it was fixed. If anyone knows who is running that server, send them a message and tell them they need to restart it. after that, it'll be all good again.
Yeah, I also think it could be a server issue. We had that with our servers a few times. A restart will usually help, but it could be also a hardware problem with the server.
Well seeing as the network connection to the server is dropped when you click on "leave" but the client gets stuck I'd think that the client is probably expecting some sort of event or something... not sure what exactly is the "proper" disconnection procedure for LFS... in which case some sort of check could be made (LFS .oO "I'm in the multiplayer raceprep menu but I'm not connected to a server - oops!").
D'oh... right you are...
Posts moved to correct thread. Thanks for pointing that out Bean0
It's true that this bug is worse in U28, because the close button now does a safe exit, and that was impossible because the server "bug" causes your LFS never to check for a broken connection.
Anyway, I have now fixed it version so it doesn't get stuck any more. LFS automatically exits after about 15 seconds. You will have the fix in U29.
Pasci's report also gave me a host to check and I've now found out what the problem is with that kind of host.
EDIT :
I've now managed to update LFS so it can join that host and stay connected. I think the situation is that all hosts are on a 49.7 day cycle. During the second half of that cycle, you can't join the host, with your current version. After 49.7 days, the cycle restarts and the host will be ok again.
If everyone had U29, they could all join that host. But I think something will probably go wrong after 49.7 days, when the cycle restarts. Something like, everyone would get kicked out then the host would continue as normal. But I'm not sure yet.
Anyway, the new fix will mean that hosts are ok for 49.7 days, which is double the current limit. The case is not closed...
Anyway, I have now fixed it version so it doesn't get stuck any more. LFS automatically exits after about 15 seconds. You will have the fix in U29.
Pasci's report also gave me a host to check and I've now found out what the problem is with that kind of host.
EDIT :
I've now managed to update LFS so it can join that host and stay connected. I think the situation is that all hosts are on a 49.7 day cycle. During the second half of that cycle, you can't join the host, with your current version. After 49.7 days, the cycle restarts and the host will be ok again.
If everyone had U29, they could all join that host. But I think something will probably go wrong after 49.7 days, when the cycle restarts. Something like, everyone would get kicked out then the host would continue as normal. But I'm not sure yet.
Anyway, the new fix will mean that hosts are ok for 49.7 days, which is double the current limit. The case is not closed...
just out of curiosity...
Scawen, you are talking about this 49.7 day cycle. I understood it that way that every lfs server (from u29 on?) does an restart at the end of this cycle. Is there any special reason for this?
Scawen, you are talking about this 49.7 day cycle. I understood it that way that every lfs server (from u29 on?) does an restart at the end of this cycle. Is there any special reason for this?
No, it doesn't restart.
It's just that the number of milliseconds since the program started is stored in a 32 bit integer, that can only go as high as 4294967296. And that is 49.7 days.
After 49.7 days, the clock suddenly goes back in time 49.7 days, instead of forwards, with unknown consequences. The result of those consequences is the host hangs and can't be joined. So it's not even a cycle, in fact.
Your version can't even join any host which is in the second half of that cycle (Team Eire Server). My version can join at any time, but the host still hangs after 49.7 days and says "Did not receive guest info" if I try to join.
I am going to fix this today.
It's just that the number of milliseconds since the program started is stored in a 32 bit integer, that can only go as high as 4294967296. And that is 49.7 days.
After 49.7 days, the clock suddenly goes back in time 49.7 days, instead of forwards, with unknown consequences. The result of those consequences is the host hangs and can't be joined. So it's not even a cycle, in fact.
Your version can't even join any host which is in the second half of that cycle (Team Eire Server). My version can join at any time, but the host still hangs after 49.7 days and says "Did not receive guest info" if I try to join.
I am going to fix this today.
Scawen, Thanks for looking at the issue. Just updated to U30 and I'm going to run some more tests.
Up till recently I've had a restart script running each morning. So I'll disable and report back with the results.
Up till recently I've had a restart script running each morning. So I'll disable and report back with the results.
I look forward to hearing the results.
How often would you have expected a problem to occur in the past, if your restart script had not been running?
Couldn't say exactly, but once or twice a week. I know one thing for sure, it wasn't consistant like a timer varible rapping, hence the need to run a nightly script just to catch it, if it was hung.
The only thing I'm doing out of the norm is running 5 instances from a shared directory, each with their own config. They run as a (SYSTEM) service (connected to the console session.) A couple run the RACE MANAGER insim app also.
Anyhow I'll keep you posted on the results. Should it go a miss again, I'll leave it in the state for you to investigate. If necessary I can arrange access to the box if it helps.
Rob
The only thing I'm doing out of the norm is running 5 instances from a shared directory, each with their own config. They run as a (SYSTEM) service (connected to the console session.) A couple run the RACE MANAGER insim app also.
Anyhow I'll keep you posted on the results. Should it go a miss again, I'll leave it in the state for you to investigate. If necessary I can arrange access to the box if it helps.
Rob
Looks good so far. After removing the daily restarts, I haven't yet seen a it happen again. Still monitoring and will update again in a week or so.
Looks to be fixed. Thanks Scawen.
1
2
stuck in selection page [ and can't leave empty server ]
(38 posts, started )
FGED GREDG RDFGDR GSFDG