Players kicked "Server Full" when server is not full.

Not sure why you don't have logs. Going to need logs at a minimum, and some other detailed information would be useful.

 

RTFM

 

I've already spent dozens of unpaid man hours troubleshooting this including reinstalling several times. I'm notifying FP of a bug to help them with their commercial product. I don't care to spend another 30-60 minutes sanitising a log just for the post to disappear w/o comment like other posts I've made. This problem should be trivial for FP to verify and it's well known enough that Smegz0r's already working on his own solution.

 

What do you need to "sanitize"?   You have questionable code in your mods that wouldn't be acceptable?

 

If you don't want to actually submit a report, then don't.  If you do, then actually do it. Provide the required information.

LOL no. All we have are Allocs and botman. By sanitise, I mean do basic due diligence to ensure there's no personal information, sensitive information, passwords whatever in the log segment.  If it were a short excerpt of a minute or three either side of an incident then sure, no problem but here we're potentially sending a hour or two's worth of logs from our prime time.

6 hours ago, trevorjd said:

LOL no. All we have are Allocs and botman. By sanitise, I mean do basic due diligence to ensure there's no personal information, sensitive information, passwords whatever in the log segment.  If it were a short excerpt of a minute or three either side of an incident then sure, no problem but here we're potentially sending a hour or two's worth of logs from our prime time.

There is no PI in the files. The server IP is obfusticated , and player information isn't PI.

 

Send the full log, don't edit it, and note the times when it happens.

It's great to get that assurance but I still need to check it myself. We've bumped the slots to 30 and things have been quieter in the last couple of days so our 6 daily reboots are keeping up with it but I'll keep an eye out if/when it happens again.

We've had less players since xmas so it's been hard to get a sample of the problem. We got one that looked like it might be the same this evening but I wasn't able to reproduce it personally and the user in question didn't try again. There were only 13 or 14 of 16 online at the time.

 

Incident was at 7:07 when player snu2 was kicked due to server being full.

 

https://www.oldfarks.com.au/funpimps/output_log.txt

 

It was happening a lot from when A20 went from experimental to public up until christmas day. We had lots of players coming and going and after a couple of hours more and more people would report this problem. We worked around it by putting scheduled reboots every 4 hours. We don't back up the log files so I can't provide those. I would expect that a stress test setup simulating ~30 players trying to connect/disconnect should be able to trigger it after a few hours.

Good news! I think we've found the problem.

 

It seems that TCAdmin enforces a connection limit, even if a game doesn't. When you set the slot count via TCAdmin webUI it updates the configfile. However if you later change the conf (as this old-school IT guy is inclined to do), TCAdmin's doesn't know about it. If TCAdmin is set to a lower slot count than 7D2D, it will kick players when it hits that number, even if the server is happy to let them in. In hindsight, that should've been a clue that it was a Kick with a reason of (Server Full). We tested this with a few values and were able to replicate it reliably. Since ensuring both TCAdmin and the conf are on the same page, we haven't had the problem.