Warning: Can't synchronize with repository "(default)" (Unsupported version control system "svn": No module named svn). Look in the Trac log for more information.

Ticket #1460 (closed defect: worksforme)

Opened 12 years ago

Last modified 11 years ago

TG crashes after a random amount of time not related to mysql connection

Reported by: gamegiants Owned by: anonymous
Priority: high Milestone:
Component: CherryPy Version: 1.0.3
Severity: normal Keywords:


Upgraded from 1.0.1 to TurboGears- TG crashes at random times when run as:

nohup ./start-ship.py > /dev/null 2>&1&

Used to have a similar problem under 1.0.1 related to losing the db connection but added a script to load the application every 15min via a cron job (this bug still not resolved in should be a stopper!) to make sure the db connection is "pinged".

Application are now crashing in even if the application does not use the db (does not even open a connection) or if it is "pinged" as descibed above

Change History

comment:1 Changed 12 years ago by gamegiants

Here is what it looks like when it goes down: - - "GET /cat HTTP/1.1" 200 5516 "" "Mozilla/5.0 (compatible; Yahoo! Slurp;  http://help.yahoo.com/help/us/ysearch/slurp)" 2007-08-01 20:00:02,125 cherrypy.msg INFO ENGINE: SystemExit? raised: shutting down autoreloader 2007-08-01 20:00:02,132 cherrypy.msg INFO HTTP: HTTP Server shut down 2007-08-01 20:00:02,147 cherrypy.msg INFO ENGINE: CherryPy? shut down

Right after getting crawled in this case....

comment:2 Changed 12 years ago by faide

Do you have any more info to give us ? I have one such case of system exit raised, but I know it comes from a Java library I use through jpype... in some special cases it brings down the entire python process without anythin I could do.

Do you have any special methods, libs or processes that could make the system exit happen... Do you monitor the resources on your machine ? I had a case where a server hosted on an OpenBSD server was starved for threads and just crashed badly...

Best regards and thanks for reporting this bug.


comment:3 Changed 11 years ago by faide

  • Status changed from new to closed
  • Resolution set to worksforme

I have not been able to reproduce this bug. I have tg production applications running 24/24 on the web with users and attackers hammering them and they are rock solid.

I'll close this ticket for the moment. Please reopen with some more information like a traceback and server logs.

comment:4 Changed 11 years ago by kvdb

In our production environment (TG 1.0.4b2, SA 0.4.3), we also experience frequent crashes.

Some of the crash reports from the python process: * glibc detected * double free or corruption (!prev): 0x0000000002124cb0 * * glibc detected * double free or corruption (out): 0x00002aaaae6f8d40 * * glibc detected * double free or corruption (fasttop): 0x00002aaaaead3d70 *

We think it may be related to SA. But we're unsure.  http://www.mail-archive.com/sqlalchemy-users@lists.sourceforge.net/msg01783.html

Hope any of this information is helpful.

comment:5 Changed 11 years ago by faide

Are you by any chance using cx_oracle ? Kvdb could we continue this discussion in the ML maybe ?

Note: See TracTickets for help on using tickets.