Server Help

[i88gerbils]

Code: Show/Hide bash$ ./bin/asss asss 1.3.2 built at Oct  7 2004 09:33:00 Loading modules... I <cmod> loading C module 'unixsignal' from 'internal' I <cmod> loading C module 'deadlock' from 'internal' I <cmod> loading C module 'prng' from 'internal' I <cmod> loading C module 'logman' from 'internal' E <deadlock> deadlock detected, aborting Aborted

I am not sure what is wrong. Anything I can try to figure that out?
_________________
Oldbie Server Help

[Dr Brain]

Try removing the persist module from modules.conf (and whatever is dependent on it).

See if that changes anything.
_________________
Hyperspace Owner

Smong> so long as 99% deaths feel lame it will always be hyperspace to me

[Grelminar]

The deadlock thing just means something is blocking the main thread for 10 seconds.

If it's really persist that's causing the problem (it only happens on some machines), the best way I've found to fix it so far is "rm -f data/__*" before starting the server. That clears out the bdb log files, which sometimes confuse it after an unclean shutdown.

On the other hand, if it's not persist, it might be taking a really long time to do dns lookups for directory servers. Try disabling directory, or manually forcing a dns lookup on your directory servers before starting the server (they'll be cached and it shouldn't take any time to resolve when the server queries for them).

[Smong]

Did you fix the thing where a debug build made it deadlock on starting up?

[i88gerbils]

Ah. dnsing directory servers was the problem. If you put the ip address of the dir. serv instead will it still try a lookup?

[Grelminar]

I think an ip should resolve instantly to itself. But I have a better fix: make deadlock wait 60 seconds when it starts up, before it aborts after 10.

I think that's better than just having it not do anything at all until all of modules.conf is loaded, because it would be nice to detect deadlocks during module load. Not that I'd expect there to actually be one, though.