Topic: Server becomes unresponsive with wsgi application

[andyhasit]

I have Apache 2.2 installed on Windows to host a web2py application using wsgi.

Apache crashed once stating it had run out of worker threads, so I inscreased ThreadsPerChild to 500.
It has since crashed/become unresponsive without giving anything in the error log.

I set up mod_status, and am looking for help understanding what I see when looking at server-status page.

It says 135 requests currently being processed, 365 idle worker. However I don't think there are 135 people connected. Here's the request lines:

Srv PID Acc M SS Req Conn Child Slot Client VHost Request
0-12 5384 0/0/0 W 1384954483 0 0.0 0.00 0.00
... the line above is repeated identically 125 times
0-12 5384 0/32/32 W 0 0 0.0 0.46 0.46 127.0.0.1 dalgety-svr8.havelock.int GET /server-status HTTP/1.1
0-12 5384 0/0/0 W 524 0 0.0 0.00 0.00 192.168.202.109 dalgety-svr8.havelock.int GET /HubForms/ProjectFolderIndexing/index HTTP/1.1
0-12 5384 0/0/0 W 524 0 0.0 0.00 0.00 192.168.202.109 dalgety-svr8.havelock.int GET /HubForms/capacity_requests/index HTTP/1.1
0-12 5384 0/0/0 W 524 0 0.0 0.00 0.00 192.168.202.109 dalgety-svr8.havelock.int GET /HubForms/Timesheets/index HTTP/1.1
0-12 5384 1/3/3 W 575 0 46.8 0.05 0.05 192.168.202.109 dalgety-svr8.havelock.int GET /HubForms/Timesheets/none HTTP/1.1
0-12 5384 0/5/5 W 556 0 0.0 0.01 0.01 192.168.202.109 dalgety-svr8.havelock.int GET /HubForms/capacity_requests/index HTTP/1.1
0-12 5384 0/121/121 W 564 0 0.0 1.64 1.64 192.168.202.109 dalgety-svr8.havelock.int GET /HubForms/capacity_requests/index HTTP/1.1
0-12 5384 0/5/5 W 561 0 0.0 0.06 0.06 192.168.202.109 dalgety-svr8.havelock.int GET /HubForms/capacity_requests/index HTTP/1.1
0-12 5384 0/3/3 W 575 0 0.0 0.08 0.08 192.168.202.109 dalgety-svr8.havelock.int GET /HubForms/capacity_requests/index HTTP/1.1
0-12 5384 0/5/5 W 575 0 0.0 0.03 0.03 192.168.202.109 dalgety-svr8.havelock.int GET /HubForms/ProjectFolderIndexing/index HTTP/1.1

Question I'm asking myself and need help with:
1) What are those 125 requests, and why are they not exiting. Are they anything to worry about? why is there no Client/VHost/Request? Could they be from the python process? Could they be what's causing the issue?
2) The Acc entry 0/121/121 on /HubForms/capacity_requests/index - is this representative of an issue?
3) When I restart Apache, it typically says "Terminating 375 threads that failed to exit" but that number doesn't match up with anything. Are these the idle threads?
Also I'm not sure what "restart" actually means, i.e. what stops and what stays (i.e. why am I on generation 0-12?).
4) Is any of this anything to do with why Apache becomes unresponsive, or should I be looking elsewhere?

Further info: I am a newb with regards to web & server stuff. This is only deployed on an intranet. There is no SSL. No errors in event manager, or web2py's error log. CPU usage seems low & stable (but didn't check at point of crash). I suspect some ajax calls in parts of the code may be a bit iffy. The web2py application was formerly on rocket webserver, where it had similar issue of becoming unresponsive, so we moved it to Apache and it worked fine (until now!).

[James Blond]

You should examinate your access log and used "combined" format.