Server Help

Cyan~Fire · Posted: Sat Sep 24, 2005 9:57 pm Post subject: Laggy shutdown on Windows

Since this is a compile-out-of-the-box, I assume other people have this problem. ASSS actually terminates 40-50s after I hit ctrl+C.

This most in-depth info I've found using gdb is this (I'm no expert), captured somewhere around 10s after ctrl+C:

Code: Show/Hide

* 3 thread 1100.0x3d0 0x77fa018d in _libwsock32_a_iname ()
2 thread 1100.0x70 0x61186380 in ptw32_cancel_self ()
1 thread 1100.0x78 0x77f82829 in _libwsock32_a_iname ()

I don't know if this means anything to anybody, if you want me to do sometime else, tell me.

I wish I could get Dev-C++ to use the right working directory.
_________________
This help is informational only. No representation is made or warranty given as to its content. User assumes all risk of use. Cyan~Fire assumes no responsibility for any loss or delay resulting from such use.
Wise men STILL seek Him.

Mr Ekted · Movie Geek Gender: Joined: Feb 09 2004 Posts: 1379 Offline

kill -9
_________________
4,691 irradiated haggis!

Bak · Posted: Sun Sep 25, 2005 2:47 am Post subject:

remove the deadlock module and try again
_________________
SubSpace Discretion: A Third Generation SubSpace Client

Cyan~Fire · Posted: Sun Sep 25, 2005 10:40 am Post subject:

No, it's not only deadlock because it waits much longer than 10s. This is new in 1.4.0.

Cyan~Fire · Posted: Sun Sep 25, 2005 11:38 am Post subject:

Ahhh, I've got it. I was shutting down before a minute was up, so deadlock was finishing that minute before me. Net also waits 10s on select, but that's not too big of a deal.

Mr Ekted · Movie Geek Gender: Joined: Feb 09 2004 Posts: 1379 Offline

Cyan~Fire · Posted: Sun Sep 25, 2005 8:42 pm Post subject:

Well the deadlock module is the one waiting 60s, and that's only once. If I'm understanding the code properly (it's using the pthread library about which I know nothing), it should cancel immediately on unloading the module, but it seems that's not happening. I can live with a 10s delay from select(), though.

Dr Brain · Posted: Sun Sep 25, 2005 8:49 pm Post subject:

I've never had any issue with ASSS not shutting down immediatly on windows. The problem is either something introduced in a version more recent that what I last used, or a pecularity in your compiling enviroment.
_________________
Hyperspace Owner

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

Cyan~Fire · Posted: Mon Sep 26, 2005 4:34 pm Post subject:

It happens on the pre-compiled build too, so it's either the version or the system. Did you use the deadlock module?

Dr Brain · Posted: Mon Sep 26, 2005 5:13 pm Post subject:

I don't think deadlock compiled for me, so probably not.

But I never had the problem with net, either.

Cyan~Fire · Posted: Tue Sep 27, 2005 7:07 pm Post subject:

Possibly you were using ?shutdown instead of Ctrl+C, and therefore the command receipt ended the recv thread immediately?

Edit: Cancel that. Just tested, and still the delay. Maybe I'll post it as a bug.

Grelminar · Creator of Asss Joined: Feb 26 2003 Posts: 378 Offline

I reduced the select timeout in net to 1 second.

About deadlock delaying the shutdown, that really shouldn't be happening because of the "pthread_setcanceltype(PTHREAD_CANCEL_ASYNCHRONOUS, NULL)". Reports on this thread seemed to indicate that it worked...