server uptime

[Jerry Feldman]

On Wed, 19 Mar 2008 14:32:54 -0400
Alex Hewitt <hewitt_tech at comcast.net> wrote:

> In my experience the stability of any system has to do with it's usage.
> With servers running programs that are reasonably stable up time will
> certainly be many months and can stretch into years. Any system that for
> example is running unpredictable loads such as one might find in a
> time-sharing university setting are less likely to have long uptimes.
> The bane of server operations are applications with memory leaks. If
> these apps aren't restricted that will consume all available memory and
> eventually cause the system to swap it's brains out. User space apps can
> usually be prevented from taking the system down but a memory leak in a
> service can easily make the system crash or become unavailable. 

We had an old ProLiant server running at the BLU with a 2 year+
uptime.  The system finally died and we cannibalized it. One of the
nicer things in Unix and Linux is that when you have an application
with a memory leak you can easily cycle it down, even if it is a
daemon. In Red Hat parlance service <daemon name> restart will clean up
many things especially in web servers.  Even drivers can be recycled
without causing a reboot. I don't recall the circumstance, but I
recycled a working network driver.. In this case I think I wrote a
short script to rmmod the driver and to insmod the new one. The system
remained up, and I was able to reconnect seconds later. 

-- 
--
Jerry Feldman <gaf at blu.org>
Boston Linux and Unix
PGP key id: 537C5846
PGP Key fingerprint: 3D1B 8377 A3C0 A5F2 ECBB  CA3B 4607 4319 537C 5846
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 194 bytes
Desc: not available
Url : http://mail.gnhlug.org/mailman/private/gnhlug-discuss/attachments/20080319/94744190/attachment.bin