Timothy Martin wrote:
> Now that it's self monitoring i have yet to catch it in this state --
> but right now as things are running well i have only one process. I
> have an idea that might be helpful -- to set monit to alert me when
> exim becomes unresponsive but not restart it. The problem is that it
> uses email to alert me so if email goes down i don't get an alert :P I
> have to figure out a way around that.
Have it append to a file instead of e-mailing directly.
Periodically have chron (try to) email you that file.
Chances are, as that activity is delayed in time, it will succeed.
If not, you can ssh in and read it, or have it written to a place your
http(s)d (if any) serves, and browse to it.
Either way, while it won't be 'real time', you can't *quite* have that
anyway,
even if 'camped on' a tail -f of the logs or having them streamed to
console.
'puters are way faster than real folks, as are their hiccups.
At least it will preserve the timestamp of the event so you can compare
it to
what was connected at the time, etc.
- Interested, as I have a similar odd situation of inordinate delays of
messages with attachments, even if acl checks, demime, clam and SA are
disabled. Server shows no load, connection times out without apparently
passing all the traffic, though *sometimes* the entire attachment has
come in, but the connection has remained open - eventually reports
failure to the sender.
Primary suspect is that I have damaged the handshakes somehow. Exim 4.54
(per this header).
