Posted: Sun Feb 11, 2007 6:52 am Post subject: Crash dump 097 rc2
I am getting exceptions on nearly every shutdown now. POL097 rc2
It will always crash when shutting down if the shard has not been running for very long (eg 10 minutes) but can sometimes not crash when its been running for longer.
An example of one exception:
Console entry:
checkpoint: exit signal detected
checkpoint: waiting for child threads to exit
checkpoint: child threads have shut down
checkpoint: writing data
Writing data files ........ Done! 12938 ms.
checkpoint: cleaning up vars
checkpoint: cleaning up scripts
checkpoint: cleaning up listen points
Unhandled exception! Minidump started...
################################
..
..
Last Script: can be anything - in this instance it's doshutdown.ecl - the PC value in the error points to the Shutdown(); line in the script. Can't see how the last script entry can be causing the crash.
The next line in the console was going to be unloading data, but it never printed that message.
The minidump entry is the last line in the console until restart.
script.log only contains:
[02/11 20:53:58] 0x005d984c
whatever that means. It's time is at the startup anyway so isn't relevent.
Pol.log
Quote:
scripts/www/server_stats.ecl 5313 1 5313 0.01
scripts/www/servperf/shutdown.ecl 77 2 38 0.00
scripts/start.ecl 22232 1 22232 0.03
[02/11 20:53:43] Unhandled Exception! Writing Minidump file.
This file with explanation and last lines from log files should be given to the development team.
Saved dump file to 'POL097-2007-01-04 RC2 Coregina-20070211202115-0.dmp'
Last Script: pkg/world/saver/doshutdown.ecl PC: 39
[02/11 20:53:58] Logfile opened.
Other information?
Nobody logged on. Shard had been running for around 5 minutes with cpu usage low. Shard restarts ok after closedown so the worldsave seems to be working ok
No entries in leftovers.txt or orphans.txt
Attached is minidump of a crash when doshutdown was the last command.
Despite the shutdown crash dumps I reported in the first post, I decided to cut the shard over to 097.
However we started getting assertion failures every few hours and restarts. (players are not so thrilled) This is not just at shutdown either and the darned logs don't seem to help much. When I get home from work I will have to revert to 095 with little to explain what is happening - a dozen or more crash dumps are about all I have. The test shard doesn't have any problems at all - but that's possibly because players are doing things they never tested.
What a mess. Do you want all the crash dumps? Do they tell you anything?
Even though there hasn't been a reply, I've been doing some more work on this because I've done too much work on 097 to stop trying.
I think the critical line in my report was:
The next line in the console was going to be unloading data, but it never printed that message.
I have been doing a lot more tests and come to the tentitive conclusion that pol does not close down cleanly when there are a lot of events in the npc queues.
POL will close down cleanly when there is little fighting going on in the shard.
Author
Message
MuadDib POL Developer
Joined: 13 Feb 2006 Posts: 830 Location: Indiana, USA
Posted: Sun Apr 08, 2007 11:40 pm Post subject:
The dump you posted, was when reading a weapon's speed entry (speed, not delay etc). O_O
If you still have them, post a few more of the dumps please.
I didn't keep the original ones from the same time period.
The crashes come and go and right now I can't generate one on RC2.1. Heh.
Since it crashed reading speed (cached itemdesc??) maybe the itemdesc files had already been unloaded from ram and the shutdown prevented it from reading a new set? Pure guess.
