MailScanner lock up Problem

[Glenn Steen]

On 10/09/2007, Paul Kelly :: Blacknight Solutions <paul at blacknight.ie> wrote:
> Glenn Steen wrote:
> > On 07/09/2007, Paul Kelly :: Blacknight Solutions <paul at blacknight.ie> wrote:
> >> Steve Freegard wrote:
> >>
> >>> By the looks of the strace - my gut tells me that the problem is related
> >>> to syslog.  Hopefully that might point you in the right direction.
> >>>
> >>> Also - make sure that 'nscd' isn't running as that's caused me no end of
> >>> trouble in the past with similar symptoms.
> >> Syslog you say. Hmmmm. No joy on that front. Also don't have nscd enabled.
> >>
> >> Paul
> >
> > The obvious questions are: What did you change, prior to the problems
> > starting? Do you monitor your HW with something like SMART
> > capabilities etc? Anything like that indicating a failing drive? How a
> > bout controller/RAM/etc etc?
> >
> > Unfortunately, SW have to run on HW... and HW can fail in the most ...
> > interesting... ways:-)
> >
>
> Turns out it was a rather interesting file system issue. I had to visit
> the machine in the data centre in the wee hours of Saturday morning
> (damn I hate being on-call) and perform open heart surgery on the file
> systems.
>
> We're doing a mix of Raid 1 + LVM on top of it and a power cycle of the
> machine during the previous few days caused some oddness with /var. This
> I believe was causing MS to die silently.
>
> It's been "ok" since. /me touches wood
>
Ah. Might explain things:-).
Since you will surely keep a very watchful eye on that particular
system for the next week or so ... you needn't be too superstitious...
Just keep the chicken wings (hmmm, shouldn't that be feet?:) handy,
for when a "double-kingston" isn't enough voodoo:-):-)

Cheers
-- 
-- Glenn
email: glenn < dot > steen < at > gmail < dot > com
work: glenn < dot > steen < at > ap1 < dot > se