Recommended frequency for reboot

[Jose Rafael Leal]

How often is it recommended to reboot your hp-ux server?

[Sanjay_6]

Hi,

We have some systems which have not been rebooted for quite some time. There is no hard and fast rule about rebooting.

Hope this helps.

Regds

[Eileen Millen]

It depends on what they are being used for.
Our database servers get rebooted once a month before the month end work is done.
The servers used for user logins should probably be rebooted weekly.
The compute servers are up for many months at a time.

Eileen

[Craig Rants]

A reboot is a subjective topic. My personal preference is once a quarter, you could coincide this with the install of the Bundled patch releases.

Do what works for you and your environment.

Good Luck,
C

[Uday_S_Ankolekar]

Hi,

There is no rule as such. If the system is performing well I don't see any reason to reboot.

-USA..

[Volker Borowski]

It depends....

Our SAP server was down at 31.12.1999 for Y2K change, came up again on 1.1.2000.
Actually it had a reboot two weeks ago for a bad disk (no HOT-plugs).
But beside this it has been up and up and up...

If you have an application, that has a memory-leak and no patches in sight, it may be needed to reboot every week.

Do not know if it helps
Volker

[Patrick Wallek · Accepted solution]

Oh boy!!! You probably will get more responses to this than you really wanted!

Everyone has differnet ideas about this. Some will say do it on a regular basis (like monthly maybe), others will say don't reboot unless you have to.

There are HP-UX machines out there running that have been running for years without being rebooted. If your application is written well and there are no hardware issues and critical patches that you have to install, you can leave the machine up and not worry about it.

If your application is not written well and you start having memory issues, then you may have to reboot frequently. The thing to do there is to get the application corrected!

I'm sure you'll get lots of responses to this.

Have fun reading!

[Vincent Farrugia]

Hello,

As a rule, as long as it is working, you should leave it working i.e. do not reboot it. Servers are there to be as highly-avalaible as possible after all.

However, there are some philosophies which state that you can do a server good if you reboot it once in a while. I was taught to leave the server working if it's working like a champ. (I've read this expression "like a champ" from these forums and I'm starting to like it :-) )

HTH,
Vince

[A. Clay Stephenson]

If you have a well patched system with stable apllications software, the reboot frequency should asymtotically approach zero. In real life, I have had quite a few systems up for years without a reboot. The exception to this rule is software with memory leaks - the answer there is to fix the software. Your box should really only need to come down for periodic patch installation.

[James R. Ferguson]

Hi:

This is Unix. If you do not have application's with memory leaks and you do not kill processes with 'kill -9' so that they have no chance to cleanup shared memory, then there should be little to no need to reboot unless it is for hardware maintenance or patch application which requires a reboot.

Regards!

...JRF...

[harry d brown jr]

It depends (and not the type you wear). Rebooting a server just to reboot it is akin to tossing off your main circuit breaker in your house just to reset the clocks and vcr's to a flashing 12:00 (unless you have batteries in them). I have had servers run for years without a reboot, while others need more frequent reboots because of crappy applications that like to have memory leaks.


live free or die
harry

[Darrell Allen]

Do you get the feeling there is no generally acceptable rule about how often to reboot? I like to reboot every now and again just to make sure the system will. Yes, systems generally run for long periods of time without needing to be rebooted but changes are made that could keep it from booting. Hopefully you are patching your system on some regular basis. I'd always reboot after installing a patch bundle, even if it is not automatic. I'd rather the system fail to come up in a controlled test than when I'm not testing and I need it back up NOW!.

Darrell

[Marco Paganini]

Hello,

As long as you don't have any applications leaking memory and no bugs are known related to big uptimes, you don't need to.

We run a very critical application here and we have HP servers with uptimes of over 1200 days...

Regards,
Paga

[John Bolene]

We mostly reboot only because installing some patches require it.

I had an internal DNS server that was up over 3 years that had to be moved to another location. It also had its IP changed at the same time. The location that it was installed in was being remodelled and rewired for gigabit. It was rebooted 3 years ago because the UPS system was upgraded and all building power had to be turned off.

Otherwise, the rest of the servers are only rebooted because a problem shows up, either because of a memory leak, needs a patch, or whatever.

[Thierry Poels_1]

Hi,

the answer is again "It depends!"

We have servers which are rebooted "never" (only for hardware interventions, patches, ...) and are sometimes kept up & running for a year.
On the other hand we have a server which is rebooted weekly, because of many hung (third-party ;) processes.

I think the ideas is: don't if you don't need to.

regards,
Thierry.

[James Beamish-White]

This quesiton would probably come from an experienced NT sysadmin...been there, got the BSOD t-shirt.

IMHO, the only reason to reboot a server, other than system maintenance (patching, kernel tuning, certain disk volume changes) would be to clean up zombie processes, as they are the only thing I have come across which really need a reboot to clean up, over restarting a process only.

You could run a script like:

if [ `ps -ef | grep defunct | grep -v grep | nl | tail -1 | awk '{print $1}'` -gt 20 ];then
echo "Zombie overload, think about rebooting" | mailx -s "System Admin Task" you@yourdomain
fi

to check if there are too many zombies and then choose to reboot (Note I just made that script up, and haven't checked it, so it may not work).

Cheers,
James

[Edward Alfert_2]

I think you'll like this... i know i'm a curious person and often check it...

http://uptime.netcraft.com/up/graph/?host=www.hp.com

you can check the uptime of any webserver...(and also get other information like os, webserver platform, etc.)

it seems by the above link query that the www.hp.com webserver get rebooted about once per month but that the max time it has been up is 115 days.

enjoy.

Ed
"nosey bastard"

[Frank Slootweg]

Just to make sure it is not lost in the multitude of other, equally valid, responses, I would like to stress Darrell's valid reason:

*Test* that the system *will* boot if needed, especially if you have made boot-related changes (like lvlnboot, mkboot, (boot/root/swap) mirroring, etc.).

I.e. such a reboot has nothing to do with 'clearing' things, but to make sure that the system will boot again after things like a failure, patch installation, etc.. Recent threads in this forum indicate that that can be a wise precaution.