"George Sakkis" <ge...ail.com> wrote:
> Didn't have much luck with this in the Pyro mailing list so I am
> trying here, just in case. I have a Pyro server running as a daemon
> process and occasionally (typically after several hours or days of
> uptime) a ConnectionClosedError is raised when a client calls a remote
> method. Both client and server run on the same machine, so I don't
> think it's an actual network connectivity issue. The problem is that
> once this happens, it is persistent; retrying to call a few times does
> not work although the server process is still alive. The current
> "solution" is to manually kill the server and restart it but obviously
> this is not ideal. Is there a way to either prevent or at least
> recover automatically the server when it hangs ?
If restarting the server sorts it, why don't you run the server as a
subprocess in a higher level script, and exit with an error code
if the error strikes?
something like:
While True:
error = os.system('python current_server_name.py')
if error:
print 'Restarting the server - error returned was:',error
continue
else:
break
It would obviously need a try - except in the existing
code to catch the exception, and some jiggery-pokery
if the code is threaded to ensure a "clean error exit",
but apart from that, I can't think of a reason for it not
to work.
It may be that the reason for your hang up is also a
threading issue - it could be that the server is alive but
that some critical thread has been killed by the error.
But here I am guessing...
- Hendrik
--
http://mail.python.org/mailman/listinfo/python-list
