Setting TCP keepalive for Cyrus daemons

[Bron Gondwana]

On Sat, Feb 13, 2010 at 08:58:50PM -0500, Wesley Craig wrote:
> On 13 Feb 2010, at 09:50, Brian Awood wrote:
> > I've had a patch against lmptd in bugzilla for a while;
> > https://bugzilla.andrew.cmu.edu/show_bug.cgi?id=3158
> > I think the code probably just needs to be audited, to find the  
> > places where
> > setting a read timeout on the stream would be sufficient.
> 
> In fact, configuring TCP keep alives simply masks these sorts of bugs.

TCP keepalives solve a different problem.  Sure you can do read timeouts,
but in the sync_client case, you could _theoretically_ have something
blocking on a cyr_expire on a large folder of something that slows
replication for a while - a read timeout will drop that connection even
though the remote host is still "responsive".  On the other hand, a TCP
keepalive will tell if the remote host has actually gone away.

Bron.