server outage - which logs are relevant?

[Dave Stevens]

On Tuesday, January 15, 2013 10:37:30 AM Hal Burgiss wrote:
> On Tue, Jan 15, 2013 at 12:21 PM, Dave Stevens <geek at uniserve.com> wrote:
> > > What is unresponsive about the server?
> > 
> > doesn't serve web pages
> 
> Well, I can't see why the apache logs would be useful?
> 
> I would check /var/log/syslog and /var/log/kern.log. If there is nothing
> useful there, and you don't have any special process accounting type stuff
> set up, you probably won't have any useful logging.
> 
> My experience with many, many of these is that more often than not useful
> logging does not happen. Why? A crash is often because something unexpected
> happened, and its hard to log stuff you don't expect and don't know will
> happen in the future. And once the bad sh*t happens, then its too late for
> logging.  And if its that sporadic, it will be really hard to know if
> whatever you try, is actually going to solve it. I find my systems will
> occasionally spit out a bunch of memory errors just before a crash, but its
> really difficult to tell what caused it. Or whether it might even be
> hardware faults of the host system.
> 
> My personal suggestion is to see if whatever software is being used to
> manage the VPS has any remote access stuff. I am on rackspace and they use
> openstack. I have scripts run from cron that check mine, and after some
> false positive checking, the script will remote reboot the system that is
> not responding. Not ideal, but it helps eliminate the several hours of
> downtime aspect.

OK, I'll look into that. Thanks!

D

-- 
Politics is the Entertainment Division of the military industrial complex.

~Frank Zappa