100% CPU problem

[GoogleCodeExporter]

1.Run a server 
2.Keep it running for days, have multiple players join it and play games
3.100 % CPU bug occurs after a some time.

http://harddrop.com/forums/index.php?s=&showtopic=2035&view=findpost&p=23842

I asked Blink to turn on the debug log to understand what is happening.

Original issue reported on code.google.com by olivier....@gmail.com on 10 Aug 2010 at 11:15

[GoogleCodeExporter]

I have logs from Blink's server running at 100% cpu time with debug option on. 
I don't see anything unusual there, but if you want to look on them I will 
email them (I don't want to post them in public because they contain tripcode 
passwords).

Original comment by w.kowa...@gmail.com on 8 Sep 2010 at 9:34

[GoogleCodeExporter]

well i was thinking with the debug log we would be able to see an infinite loop 
but if you dont see it then i dont know.

Original comment by olivier....@gmail.com on 9 Sep 2010 at 7:35

[GoogleCodeExporter]

can you send me the logs by e-mail and tell me when was the approximate time of 
the problem i'll look if i find anything relevant .

Original comment by olivier....@gmail.com on 11 Sep 2010 at 2:30

[GoogleCodeExporter]

Blink reports that setting log4j level to OFF fixed issue. I don't know how 
this is possible, bug in log4j? Maybe we should update budled log4j to version 
1.2.16 and see if this will help.

Original comment by w.kowa...@gmail.com on 12 Oct 2010 at 6:48

[GoogleCodeExporter]

actually he is not yet sure if this was related to problem, server worked fine 
for about 3 weeks and he got 100% usage again.

Original comment by w.kowa...@gmail.com on 26 Oct 2010 at 10:00

[GoogleCodeExporter]

This is still a critical issue, which needs further investigation.
I'd say the priority on this should be high.

I brought the new 7.4 server up today. There were quite a number of users 
connecting and disconnecting over a period of about half a day, since many 
people wanted to try out the new version (about 100 rooms created). After 
~11h50m the process hit 100% cpu usage. (total cpu time usage before was only 
~40 sec)

Original comment by bob.ins...@gmail.com on 30 Oct 2010 at 2:57

[GoogleCodeExporter]

now with admin tools we can add diagnostic tools easily to hunt down this bug. 
for example we can add admin command to print getAllStackTraces.

Original comment by w.kowa...@gmail.com on 30 Oct 2010 at 3:14

[GoogleCodeExporter]

I can reliably cause this problem to occur.

1. Start netserver, default port 9200
2. Connect to server with nc
   > nc localhost 9200
3. Close the nc connection
4. Server grows to 100% cpu usage

While the nc connection to the server is running, everything is fine. Only 
after I close the connection does the server go to 100% cpu.

If there are any settings or log files anyone wants to take a look at let me 
know.

Original comment by Colin.Bl...@gmail.com on 2 Nov 2010 at 10:39

[GoogleCodeExporter]

thanks, I was able to reproduce this problem in 7.4.0, but looks like it was 
already fixed in r521 .

Original comment by w.kowa...@gmail.com on 2 Nov 2010 at 11:29

• Changed state: Fixed

[GoogleCodeExporter]

Original comment by w.kowa...@gmail.com on 21 Jan 2011 at 4:34

• Changed state: Verified