This sounds like something we were having lately, namely that older versions of
Cyrus don't have any TCP timeouts in use, so if the other end of a connection,
typically a mobile device, goes away without closing the connection, the
connection just sits there and occupies the server until the end of time. There
was an option introduced somewhere along the 2.4 line to prevent this kind of
behaviour - i.e., an option to define a TCP timeout.
You can simply kill the ancient imapd processes. As the other end of the
connection has gone away, your users won't notice anything. The master process
will spawn new imapd processes as needed.
On Wed, May 09, 2012 at 07:42:34PM -0400, Brian Capouch wrote:
> We are running Cyrus 2.2.13, and I recently noticed that we have many
> imapd processes on the server, dating back to the day that the server
> was last booted, which was over 3 months ago.
> The entries for many of them in /var/imap/proc look like this:
> > 249.sub-174-253-10.myvzw.com [184.108.40.206]
> Note there's no user name with them.
> I have googled myself silly, and can't find any evidence that others
> have seen this problem.
> I know they need to be killed; short of rebooting the server, is there
> anything I could/should do?
> I inherited this installation, btw, and don't really have anyone to talk
> to about it.
> Cyrus Home Page: http://www.cyrusimap.org/ > List Archives/Info: http://lists.andrew.cmu.edu/pipermail/info-cyrus/