Hello Hosts.

A note about the U32 test patch - there are some important fixes in the multiplayer code, related to OOS checking. There are some situations in which Synchronisation could not be checked for a long time in recent patches, and also an old bug where several innocent guests could get kicked at the same time for a "Sync Late" usually related to someone else who was lagging.

As official Patch V will be out in just a few days, it would be great if you could use the U32 dedicated host if possible. It should be more reliable than any previous version of LFS.

Thanks!

http://www.lfsforum.net/showthread.php?t=9869

U32 patch added here. so should this eliminate most mass disconnects.. in theory?

Dead Men Racing patched to U32 yesterday evening.

Will it be possible in a future release to set the dedicated host to autoupdate (including test patches)?

We've launched an S1 version of the STCC New & Bronze server.

We did notice that at first the server was only showing when S1 was selected on the client, and wasn't showing in the list at all in S2 mode.

We've fixed it by adjusting the number of players to S1 numbers in the config, but I think given the inconsistency, it's a bug.

Sadly, S1 racers who become completely addicted to earning their STCC licence will not only have to race lots to gain points, but will also have to spend £12 more on their LFS licence in order to progress to the Silver licence server

Quote from franky500 :

U32 patch added here. so should this eliminate most mass disconnects.. in theory?

I don't know - if most mass disconnects were of the "Sync Late" type then there is a possibility that this patch will help - that is what it's intended to do but I was unable to reproduce the bug, I could only find a theoretical reason for it and much reduced that possibility (I think).

The reason I fixed that is that I was online recently and half the racers there got kicked with a Sync Late message, which coincided with someone joining and lagging (I don't know if the joining had anythimg to do with it, or if to start lagging some time after joining would have the same effect).

Only thing is, I have heard recently of mass disconnects but I don't remember anyone reporting that there was a "Sync Late" message for every person who got kicked. And that surprises me because I would expect someone reporting a bug, to report the error messages that go with it.

But maybe they did report that, I missed it and can't remember.

In short - there is definitely one type of disconnect where someone saw the error messages (me) and that one is theoretically improved (though not tested in any kind of definite way).

Quote from SamH :

We've launched an S1 version of the STCC New & Bronze server.

Thanks!

Quote from SamH :

We did notice that at first the server was only showing when S1 was selected on the client, and wasn't showing in the list at all in S2 mode.

We've fixed it by adjusting the number of players to S1 numbers in the config, but I think given the inconsistency, it's a bug.

I don't really understand what you mean. Please can you explain this one, exactly how to reproduce it. If you can simply explain step by step how to make the bug happen and how to see the bug (this bit I think understand - under certain circumstances, S2 can't see S1 hosts) that would be great.

The /maxguests and /carsmax parameters in our S1 config file were set to S2 numbers - 23 & 20 (IIRC).

This made the server visible when choosing S1 mode and browsing multiplayer games. However, it meant that the host was invisible when in S2 mode, browsing games. We fixed it by reducing the number of guests and cars to S1 numbers (15?) - Becky edited and FTP'd the config, while I closed and relaunched the server through RDC. After relaunch, the server became visible in S2 mode as well as S1.

We did test to see if the server had launched poorly first, by closing and relaunching before editing the config.. and the problem persisted until we made the edit and relaunched.

OK Sam, at first I could not reproduce this - your method didn't work, I was able to see my "too many players" S1 host in a S2 list of games - as expected because the players are limited regardless of your config file.

However, by chance I was able to reproduce the bug by a completely different method. If I issued a /pass xxx command, this sends an updated status to the master server and a small master server bug meant that the host now appeared as "host is a different version" in the S2 list of games. That is now fixed and the master is restarted with the fix (and I can't reproduce it any more). Code searching I could not find any other way this bug could come up.

In short, I could make S1 hosts vanish from list of games with 100% certainty by typing /pass xxx in the S1 host. But number of players didn't affect it in any way. Could it be you has used the /pass command and that coincided with your setting of too many players?

I just spoke with Becky, and (while we're not sure, absolutely) we think we probably temporarily passworded the server while getting it up and running.. and so there's a great big fat possibility that it's the same bug you've identified. We pinned the wrong cause on it. We were both definitely listing servers with no filters on.. we did both make sure of that.

I think I hear the sound of a bug splat

[edit] We did see a single "wrong version" in grey at the top of the list. We assumed that was ours.

Hi

I have upgraded [CD] ConeDodgersRacing 1, 2 and 3 to U32 this morning and several racers, including myself are getting a warning.

"Your skin cannot be found at www.lfsworld.net"

Yet the skin can be seen by other players and is present on LFSW

Several have tried uploading the skin again but it make no difference

The message comes up after few seconds when you go on track, however it does not seem to be everyone.

I have also put this in the patch post section

i know we are on patch V now bu tjust got a big Sync Late. so obviously not completely sorted .

We have just had a mass-disconnect on a full server using V so its still not right with 23 connections

Birder, were there any error messages when there was a mass disconnection?

I need much better descriptions, if I am to ever improve anything. I need the number of racers that were disconnected, the number of racers that were not disconnected. What was happening at the time (e.g. someone joined, someone left, someone was lagging badly). Also need to know any error messages that were displayed.

I need to see some kind of a pattern, in order to try and deduce what could be the problem.

Quote from Scawen :

Birder, were there any error messages when there was a mass disconnection?

I need much better descriptions, if I am to ever improve anything. I need the number of racers that were disconnected, the number of racers that were not disconnected. What was happening at the time (e.g. someone joined, someone left, someone was lagging badly). Also need to know any error messages that were displayed.

I need to see some kind of a pattern, in order to try and deduce what could be the problem.

Ignore my one. I just realised it was still on U20. sorry.

Quote from Scawen :

Birder, were there any error messages when there was a mass disconnection?

I need much better descriptions, if I am to ever improve anything. I need the number of racers that were disconnected, the number of racers that were not disconnected. What was happening at the time (e.g. someone joined, someone left, someone was lagging badly). Also need to know any error messages that were displayed.

I need to see some kind of a pattern, in order to try and deduce what could be the problem.

I am not sure if this was not something to do with the server company and not LFS. Franky500 has told me they had a problem at about that time, but we cannot be sure.

In the previous post he says it was on U20, but it was showing as V in the master server so i am not sure he talking about the same one.

All 23 were disconnected with no message. If it happens again i will post the full details.

Thanks, Happy new year

Quote from franky500 :

Ignore my one. I just realised it was still on U20. sorry.

If you are talking about the mass-disconnect on [CD] 1 it was on V not U20.

Quote from birder :

If you are talking about the mass-disconnect on [CD] 1 it was on V not U20.

I had a sync late on the 29th on a diff server, Thought it was on patch V but was on U20.

Quote from birder :

All 23 were disconnected with no message. If it happens again i will post the full details.

Thanks, Happy new year

OK, there must have been at least *some* message for each person who got disconnected. Like "X disconnected" or "Lost connection to X" or "X timed out".

Do you have logging switched on, on the host? "Log Messages" in cfg.txt - can be set to 1.

Happy new year!

We've not had a mass disconnect on the STCC servers since upgrading to U32, and then more recently going to V. Importantly, we also changed server location at about the same time as going to U32.

It's very much become apparent recently to us at UKCT that there are two types of hosting.. server hosting, and game server-oriented hosting. After months of awful problems, we realised this is the one thing that we changed (by not looking specifically for game oriented providers), fixed it by researching providers differently, and we're rolling again with brilliant data reliability and miniscule network latency.

We have been getting one error, which results in usually 2 people disconnecting.. the message says "MP Replay out of stash space". The server recovers with the message "MP Replay stash recovered", at which point two people invariably disconnect. I can only assume that one is a victim of the other.

As for mass disconnects, they're a distant memory for us right now

Quote from SamH :

We have been getting one error, which results in usually 2 people disconnecting.. the message says "MP Replay out of stash space". The server recovers with the message "MP Replay stash recovered", at which point two people invariably disconnect. I can only assume that one is a victim of the other.

The MP Replay stash space message is not really related. It's just another symptom of the lagging and causes the replay to have a slight gap in it.

The thing is, at least one player is lagging (still connected - in theory - but no packets are arriving at the host or anyone else). So your MPR saving system can't save to file, because it is waiting to process the game turn that is relying on the packets that haven't yet arrived. The time-out is quite long so eventually there are too many pospackets to keep in memory (actually I could increase that buffer size). And that's when the stash space message comes up.

Then finally the player who hasn't sent any packets to the host for a long time, gets kicked so the game can continue. It should say that X timed out, I think. At that point you don't have to wait for his game packets any more and the MPR can continue to be saved to file, and that's when you get the stash space recovered message.

The only thing a bit mysterious about what you are saying is that two people disconnect. Really, only one person should disconnect - the one who was lagging for some reason. You should have seen his network connection bar stay at the bottom while everyone else's bars rose for a while. Please keep an eye on that.

I'll keep an eye on it. I usually have Network Debug on, when I'm admining, so I'll get a collection of screenshots together (unless there's a convenient network debug log in one of the dedi directories?)

Well if you switch on logging "Log Messages" in cfg.txt.

Then all messages are logged to the deb.log file (which is deleted when you next start LFS.exe).

I've turned Log Messages on in the cfg.txt on the dedi. Is that the right place? So far, I've never personally been disco'd this way. I've only seen others.

Quote from SamH :

I've turned Log Messages on in the cfg.txt on the dedi. Is that the right place?

Yes but I hope you exited the host first, then edited the cfg, then restarted - otherwise it won't work.

Oooh.. errr... bugger!

If enough people ever leave the STCC servers we're running, for me to shut it down, I'll do that!