[lug] server memory woes (3rd attempt!)

[John Hernandez]

Michael Belanger wrote:
> Greetings everyone.
> 
> I have this new email/dns/dhcpd/nfs/jabber 
> (WayToManyCriticalThingsForOneMachine) server that has 3 gigs of RAM and 
> dual AMD Opteron processors.
>
> It stops responding to requests periodically (once a week or so).
> First, I suspected that the firewall was not allowing AMANDA to do its 
> thing (usually stops responding at night during the backups), but it has 
> still behaved the same -- even after opening up the firewall completely 
> for the amanda server.
>
> I have noticed that after a reboot it has approx 2.4 Gigs of RAM 
> available but goes down gradually to around 250 Megs available by the 
> next day.
>

Depending what tools you're using to monitor available memory, this is 
probably just normal behavior.  The kernel will generally suck up 
physical memory for things like file caching, etc.

> I tried restarting  the usual suspects like NFS and SAMBA, but the RAM 
> usage doesn't seem to recover.  Something is not freeing up Memory.
> 

Again, probably nothing alarming (for the reasons stated above).

> Other things I haven't tried restarting that could be responsible are 
> dhcpd, IMAP, and spamd.
> 
> Is there any way to effectively track that down or am I doomed to reboot 
> the server every day or so?
> 

Does it hang hard, meaning you need to press the reset button to reboot? 
  If so, it sounds like a hardware problem.  Try running memtest86 from 
the bootloader.  Be forewarned - it takes a while to run.

> The server is running Redhat 3.0 AS Release 3 x86_64 and is up to date 
> with all the redhat updates.  All applications -- except AMANDA and 
> Jabber are just the Redhat Distro RPMs.
> 
> Thanks for helping!
> 
> -M


-- 

  |  John Hernandez - NOAA Boulder NOC - 303-497-6392
  |  Mailstop R/OM62. 325 Broadway, Boulder, CO 80305
  |  PGP Public Key ID: 586A7E23