Power

Does IBM have documentation somewhere on what the proper procedure / order is for completely shutting down an IBM Power 10 and all the related hardware (HMC, external storage, etc)?  I tried doing a Google search and really couldn't find anything. 

Dear Michael

In my experience, I find this matter to be simply straightforward and not complicated in any way. A reason may be that not many people shut down all IT HWs each year.   What turns out to be more complicated and needs to be done well in many cases is how to do an orderly shutdown of the application(s) and, sometimes, OS in the server.  

For IBM i, there are a few things to know to avoid abnormally long IPL time after an unexpected abnormal end.  If you run any IBM i LPAR in your server and want to know how to do orderly shutdown of all workloads and prevent long IPL after an abnormal end, let me know.  

For powering off a Power System server, one System Profile Setting for the server in HMC named "Power off when the last logical partition is shutdown" is available for you to use it.  Its default is unchecked.

------------------------------
Education is not the learning of facts but the training of the mind to think. -- Albert Einstein.
------------------------------
Satid S.
------------------------------

Having completed a few data center migrations, and one overheat at one data center and having to shut stuff down I can see a few times when you'd have to do such operations.  (Trust me, that data center will NEVER have an overheat situation again.)

Generally you 

1. Quiesce all the applications running on each lpar of IBM i, AIX and Linux.
2. Shut down each lpar of IBM i, AIX and Linux
3. Shut down each lpar of VIOS.
4. Shut down each Power server.
5. Shut down the SAN (Ours is so new I've never done this.) Note:  Our SAN is for our Power systems only.  If you use it for other stuff you may have to move this step later.
6. Shut down the HMC.  Ours are all virtual HMC's on X86 hardware.
7. Then you start shutting down the vm servers, etc

If you regularly put patches on your lpars, vios, Power system firmware, HMC's you get pretty adept at this.  We do this quarterly.

If you are running H/A failover through some method or another then process that also.  We do switches as part of each quarterly downtime.

I do all these updates, failovers, etc remotely.  Our data centers are about 5 hours apart with our corporate office roughly in the middle.