Re: [nevow.appserver.NevowSite] Could not accept new connection (EMFILE)

On Thu, 06 Sep 2007 10:13:54 +0200, Werner Thie <[email protected]> wrote:
>Hi all
>
>I am running into a major problem when running an nevow site with LivePages 
>under low load conditions:
>
>After some time (non-determininstic), say 7-10 minutes with continous 
>interaction of say 4 clients the server disconnects the clients with no 
>apparent reason and logs a host of
>
>Could not accept new connection (EMFILE)
>
>After that, one or more tracebacks follow, which say that too many files are 
>open in /tmp. Concurrent supervising of /tmp shows, that at most one open 
>tmp file can be observed at any one time, except when the above line 'Could 
>not accept new connection (EMFILE)' is logged.
>
>uname -a
>OpenBSD sfjass 4.1 GENERIC.MP_RAID#0 i386
>
>ulimit -a
>time(cpu-seconds)    unlimited
>file(blocks)         unlimited
>coredump(blocks)     unlimited
>data(kbytes)         1048576
>stack(kbytes)        16384
>lockedmem(kbytes)    unlimited
>memory(kbytes)       unlimited
>nofiles(descriptors) 1024
>processes            532
>
>If I connect more clients (20), the problem shows up in the log as shorter 
>burst of EMFILE entries, no tmp file tracebacks, no disconnects of clients 
>at a rate of about once every three hours.
>
>Does someone out there have a clue what could trigger this problem.
>
>I am grateful for any hint, because this one could kill my project.

The traceback referring to the /tmp file indicates a problem, but the
problem isn't necessarily due to the creation of files in /tmp (although
that's sort of interesting and a bit sad and maybe we'll do something
about it).

It would be useful to know what kind of files are open when the failure
happens and if they're sockets, what state they're in.  It would also be
useful to know more about how you triggered this behavior.  Do you have
a stand-alone example which produces it?

Jean-Paul