preventive shutdown regarding HP servers

[bouazzaoui]

dear gourous,
I would like to ask you about your advice for a strange request I got from my customer :
he have rx8620 server running hpux and he have also Proliant server (ML series) and he asked me about the periodicity of shutdown for such kind of server (the min time between shutdown!)
is there any document where I could find such kind of info?
thanks for you help and patience
regards

[Dennis Handly]

What OS version are you running?
What do you think you are "preventing" by rebooting?
I assume you are thinking about patching or updating the firmware while you have it down?

[Matti_Kurkela]

With both Proliant and rx8620 servers, the minimum time between two consecutive clean shutdowns would be dominated by the time required for shutdown of application services and OS, hardware self-tests and OS + application start-up. These obviously depend on what is installed on the system. A fresh OS installation with minimum memory configuration and no applications added will certainly complete a shutdown and restart significantly faster than a system with maximum RAM and a heavily-used multi-terabyte Oracle database installed.

Unless the power is actually turned off, I don't see a reboot causing any significant extra wear and tear. At power-up, the initial power surge, the power required to spin-up all the fans and disks from standstill and the thermal expansion as various parts warm up from room temperature to their standard operating temperature would cause some extra strain - but just a reboot usually does not cause any of that.


The only reason I can think of for asking specifically the "periodicity" of shutdown would be when planning for UPS capacity and shutdown sequences.

Scenario:

Power fails and stays gone for so long that UPS monitoring programs will trigger a controlled shutdown. All systems are shut down in a controlled manner before UPS outputs are switched off, but now UPS batteries are at considerably less than full charge.

Then the power comes back up, causing systems to restart... and immediately after that, the power fails *again*.

The real question your client might be thinking:

At this point, is there still enough power in the UPS batteries to complete start-up and immediately execute another controlled shutdown after it, or will some servers crash because UPS runs out of battery power before the shutdown is complete?

MK

[Steve Reece_3]

If I understand the question correctly, it's more of a management question than a technical one.
My view is that systems ought to be rebooted periodically so that you know that the system will reboot should it be necessary. e.g. you haven't added something to the system that you need but have failed to add to the startup scripts.

A few years ago, it was necessary to reboot systems to tidy up memory leaks. These days, software is generally better and systems have more memory so this isn't so much of a problem.

Powering the kit down and then back up again will cause a small amount of wear and tear due to starting surge on power supplies etc. I've yet to meet a computer system which was damaged by shutting it down and powering it back up again.