hard crash when restarting cherrypy

I have a long-running server proxied behind Apache.
Environment is
SuSE Linux 10, Python 2.4.3, CherryPy 3.0.2. I started to
see a
problem a couple of weeks ago where the server would fail to
auto-
reload, CherryPy would crash with the error below, and I'd
have to
reboot the Linux server:

Exception in thread CPHTTPServer Thread-1:
Traceback (most recent call last):
  File "/usr/lib/python2.4/threading.py", line
442, in __bootstrap
    self.run()
  File "/usr/lib/python2.4/threading.py", line
422, in run
    self.__target(*self.__args, **self.__kwargs)
  File
"/usr/local/lib/python2.4/site-packages/cherrypy/_cpser
ver.py",
line 126, in _start_http_thread
    httpserver.start()
  File
"/usr/local/lib/python2.4/site-packages/cherrypy/wsgise
rver/
__init__.py",
line 851, in start
    raise socket.error, msg
error: (98, 'Address already in use')


I disabled auto-reload and upgraded from CherryPy 3.0.1 to
3.0.2, but
now when I try to restart the server by hand I get the same
thing, and
the only way to get CherryPy to restart is to reboot the
server.

This just started happening in the last couple of weeks, and
I haven't
changed the OS or Python distros, so it must be something in
my own
code causing this. Looking back over the changes I've made
recently a
couple of things arouse suspicion:

1. The server code has a couple of timer threads that run
every few
minutes to do cleanup tasks. Those have been there for a
while but I
did tweak one of them recently. I have code in CherryPy's
shutdown
task list to stop those threads. None of those threads have
any
sockets open directly but the server does maintain a pool of
open
PostgreSQL connections and one of the shutdown tasks is to
close all
open connections.

2. The server use subprocess.Popen to execute xmllint. It
always waits
for the xmllint process to finish. We were doing this for a
long time
in a way that appeared to be blocking the CherryPy server:

  proc = subprocess.Popen('xmllint', ...)
  result = proc.communicate(...)

I recently changed that so it would give time to other
CherryPy
threads, since xmllint can take a long time if the XML file
is big:

  proc = subprocess.Popen('xmllint', ...)
  while proc.poll() == None:
      time.sleep(0.5)
  result = proc.resultcode

Since I can't always get the server to crash or hang -- it
seems
related to how long the server has been running -- my tests
reverting
some of these changes back haven't shed much light on the
problem.

I hoping someone else has seen this, or can at least tell me
that
using subprocess.Popen or running other threads with the
threading
module or whatever else I may be doing is a likely source of
this kind
of problem. Having to restart the Linux box to update the
server code
is not going to work for us long-term.

Thanks!

Greg Jorgensen

--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the
Google Groups "cherrypy-users" group.
To post to this group, send email to cherrypy-usersgooglegroups.com
To unsubscribe from this group, send email to
cherrypy-users-unsubscribegooglegroups.com
For more options, visit this group at h
ttp://groups.google.com/group/cherrypy-users?hl=en
-~----------~----~----~----~------~----~------~--~---