LlamaServer: status and bug reports

[ano]

I see that Victoria game page has been restored. Is it possible to restore the Stasis game as well?

[Johan K]

My experience is that cron can hiccup if you get too much text output. So if you have text output from the scripts that might be the problem.

[llamabeast]

ano - sure, sorry, I'd missed that it had gone.

JK - thanks for that.

[WraithLord]

If you don't need the text output just redirect it to /dev/null
you could redirect stdout by using cmd > /dev/null, stdout and stderr by cmd >& /dev/null that is if you're uusing sh or bash. Other shells have similar concept but possibly slightly different syntax.

[llamabeast]

At the moment I've got "&> log", where log is a log file. So I guess probably that would sidestep the issue?

[vfb]

Are you running on an NFS-mounted drive?

Also what do you think of making your cron create a lock file so your server is only ever running one game at a time? You could write a little C program to exec dom to do that, and wrap the exec call with an open() of the lock file with O_CREAT+O_EXCL flags.

[llamabeast]

Er, NFS? Knowledge fail by me.

It only runs one game at a time, although I didn't quite follow that about the lock flags. It just makes a file when it starts hosting, and deletes it when it's finished, and only starts if the file doesn't already exist. There's actually like a bazillion checks and stuff in there, added incrementally to make things more robust. Still, suggestions are always welcome.

No crashes over the last 7 or 8 hostings. Could it be... working? Who can say? Presumably it will go mad as soon as I go to bed.

[Lingchih]

Llama, I'm afraid that you are going to have to stay awake and at the machine at all times. It appears it does not like it when you leave. Sounds a bit like a cat.

[archaeolept]

i think we're going to need electrodes that can be accessed from the server status pages

[vfb]

Originally Posted by llamabeast

Er, NFS? Knowledge fail by me.

It only runs one game at a time, although I didn't quite follow that about the lock flags. It just makes a file when it starts hosting, and deletes it when it's finished, and only starts if the file doesn't already exist. There's actually like a bazillion checks and stuff in there, added incrementally to make things more robust. Still, suggestions are always welcome.

No crashes over the last 7 or 8 hostings. Could it be... working? Who can say? Presumably it will go mad as soon as I go to bed.

I thought you might have network-mounted some additional drive space using NFS, since earlier you said it ran out. But if you're not sure, then it's pretty unlikely you're using it.

Using open() just makes the file existence check and creation atomic. You can probably do it from perl or php or some other scripting lang.

If you've just got a script that does

if [ ! -f mylockfile ]
then
  touch mylockfile
  do stuff
  rm mylockfile
fi

Then there is a chance two copies can execute at the same time if both do the check first before touching the lock file.

And when the first one finishes, it will remove the lock file while the other is still running, allowing another to start, etc.

But it would be a really rare problem I think, so it's unlikely to have caused what you're dealing with now.