Reboot Frequency

[Joe Pegram]

Is there a recommended reboot frequency for Win2000 Advanced Server?

I have one box that only gets rebooted when there is a problem that cannot
be fixed any other way. But we have a new box, and the primary users feel
it needs to be rebooted every week.

Apparently our software vendor (Geac) concurs with our users, citing
"...it's a common practice with most of our Customers".

Thanks.
-Joe

 

[Danny Sanders]

I have 4 Win 2k advanced servers. I only reboot them about once every month
or two.

Is there something happening with the box that is cleared up by rebooting?


hth
DDS W 2k MVP MCSE

 

[Scott Harding - MS MVP]

This is NOT common practice. Having to reboot frequently is a problem with
the applications/drivers on the system or something wrong with the OS. The
recommended reboot is NEVER!! Obviously the company software people are the
ones with the bad software and thus they tell you it is common. Typically
the need arises from software that causes memory leaks and the system slowly
gets all it memory used up to the point where a reboot is required to fix
it. I dread rebooting my servers as I have had several machines in the past
not restart because a drive finally dies on a reboot when it stops spinning.
The really is no reason other than critical updates/patches that a machine
should need a reboot. This is ALWAYS an indication of something being
written badly and some possible OS corruption etc. If there were no updates
a server theorectically should be able to run forever without a reboot. Now
I know this is not true but there is definitely not a recommended regular
reboot practice and if you are finding yourself having to do this I would
scream to hills at the software vendor who is making you do this. They need
an update for their product.

 

[Joe Pegram]

The box is only running SQL 2000, so I can't imagine how the application
failure could be fixed by a reboot. Unless resetting all the client
machines after the reboot did the fixin'.

Thanks for the reply!

-Joe

 

[Joe Pegram]

I agree! But what do I know, I'm just the programmer...

Thank you for the reply!

-Joe

 

[Hank Arnold]

This is NOT common practice.

.............................................................

I completely agree. In most cases, if re-booting resolves problems, you need
to address the problems that require the re-boot. There is, however, one
circumstance where the only dispute is whether you re-boot every night or
once a week. That is when you are running Citrix MetaFrame on the server.
There is almost universal agreement among Citrix administrators that
re-booting improves performance of the Citrix machine. The only controversy
is how long you should wait.

I dread rebooting my servers as I have had several machines in the past
not restart because a drive finally dies on a reboot when it stops spinning.

I'm not with you on this one.... I've had two drives fail on me. One finally
tanked when I re-booted the server. The other waved bye-bye just before the
office opened. I had gotten into the habit of re-starting all our servers
(only had 4 of them then) on Thursday at 6PM after the Hospice I support
closed the office. In the first drive failure, I had over 12 hours to
install a new drive and restore the files from tape. When people came into
the office the next morning, they wern't even aware that there had been a
problem. It was only when they tried to access a file that had been created
or modified after the previous backup that they saw a problem. Most were
astonished that we told them that we had a problem.The disruption was
minimal and only a couple of peolpe had any significant impact.

In the second case, we had to go through the same process, but, since we
didn't know about the problem until we got into the office, the
organization's operations were significantly impacted for almost 5 hours.

I'd rather have a hard drive failure when I'm prepared to react than at the
worst possible time....

BTW, we don't to the "all server re-boot" any more. We have software
installed that monitors the server hardware and alerts us to problems as
they start. One reason we got it was that , we could see behaviors (slow
access to network drives, network drive not connecting a boot, but
connecting when you accessed them, etc.) that, looking back were alerts that
the drives were failing.... Also, we have a lot more servers (10+), more
than one DC and a new High Availability and DR system that help minimize the
impact of down drives and systems. We still re-boot the Citrix servers at
least once every week or two, though.