[olug] Anyone defragging their linux servers?

Obi-Wan obiwan at jedi.com
Thu Oct 9 15:12:57 UTC 2008


> Thanks for the comments.  If not defragging, then do you or anyone else 
> have suggestions as to regular maintenance that should be run on CentOS 
> servers, besides running regular updates?

Just keep an eye on disk usage and keep up on patches, and you should
be OK.  That's pretty much what I did for years at BryanLGH, and it
worked well.

> I don't do updates very often either, I'm afraid of breaking something.  
> How do you guys handle this?    I could run 'yum update' on the dev 
> server, and just let everything update, and then see if the basics are 
> still working ok.  But it may be easy to miss something.  Usually I just 
> update apache/php/mysql and leave the rest alone.  Do you trust it 
> enough to just do a full yum update every night from cron?

Don't let it update unattended!  The way I did things with a couple
dozen machines at my disposal was to once a week do a wholesale update
on all the non-essential machines -- the ones which could afford to be
down for half a day while I fixed something that died (it never did).
If those machines ran well for a few days, then I'd update the critical
servers.  I'd always keep an eye on which packages were being updated,
and I'd be very careful about updating anything that I knew would
effect a service that was running on that box.  If you've got a pretty
stock box, you should be OK, but if you've done many customizations,
you need to know which updates are likely to overwrite your
customizations.

CentOS, being based on RHEL, is by nature a very stable OS.  They
don't generally make substantive changes to how things work within
each major version of CentOS/RHEL.

Despite being paranoid, the only updates that ever gave me problems
were due to my own hacks.  For instance, modifying OEM scripts or
renaming config files, both of which got overwritten by updates.

If you do your updates weekly, there'll be few enough packages in
each update that you'll be able to check each affected service after
the update to see if all is well.  Doing them quarterly or anually
makes it a much more daunting task.

Just updating Apache/PHP/MySQL and letting the rest go stale means
you're updating the dangerous stuff and skipping the easy updates.

-- 
Ben "Obi-Wan" Hollingsworth                             obiwan at jedi.com
   The stuff of earth competes for the allegiance I owe only to the
     Giver of all good things, so if I stand, let me stand on the
       promise that You will pull me through.  -- Rich Mullins



More information about the OLUG mailing list