Re: Power7/HMC: How to automatically start/stop VIOS with the hardware -- MIDRANGE-L

Good comparison. And, yes, I do have an app on one vmware instance using a
DB on another vm instance.

Here's another example.
My Power 9.
I have two vios lpars virtualizing ethernet and fiber channel for the
tape. Disk is all internal.
I have one lpar of IBM i virtualizing disk to the other lpars - a mix of
IBM i and AIX.
I'm sure it could be scripted but it would be rather tricky.
Let's see shut down the two H/A software backup lpars which will need to
communicate to the H/A software prod lpars at the other data center to let
them know they are dropping. While those are dropping bring down the other
two backup lpars supporting Domino applications, of which I normally force
a Domino replication to ensure that the databases are in sync with the
other location. While all four of these are dropping I could drop the AIX
lpar supporting IBM Storage Protect. After all five of these lpars are
down I could drop the one lpar of IBM i which virtualizes disk to all of
those five lpars. After that is down I can drop the two VIOS lpars. Then
I can drop the firmware on the Power 9.
All of this is documented and I have others follow the documentation to do
this on occasion to validate the documentation.

On Tue, Mar 26, 2024 at 10:58 AM Roberto José Etcheverry Romero <
yggdrasil.raiker@xxxxxxxxx> wrote:

I would counter with, "Why would you shutdown the hypervisor before the
VMs?" In the PowerVM world you have the hardware hypervisor, which is the
machine itself and the HMC says "are you sure you want to shutdown this?
there are LPARs running" finally you have the Virtual IO Server, which
brings services to the other LPARs.
Does your DB Server warn you about shutting down the APP server or closing
all connections before shutting it down? No, that is for whoever owns the
machine to decide.
Going back to ESXi, the storage won't complain if I tell it to shutdown
while the entire VMware environment is using it...

And that is why the whole automation ecosystem exists. You set your own
rules and use the provided tools to implement it, it is pretty easy after
all.

On Tue, 26 Mar 2024 at 11:11, Patrik Schindler <poc@xxxxxxxxxx> wrote:

Hello Rob,

thanks for having a look!

Am 26.03.2024 um 14:26 schrieb Rob Berendt <robertowenberendt@xxxxxxxxx
:

I checked my systems.
I use no "Power Controlling Partitions". This is a property you can

set

via the HMC.

There is an option on my vios lpars, under advanced properties,

"Automatic Start with Managed System". However it will not let me turn it
on.

Same here. I wonder about what is the underlying reason.

Since I frequently apply firmware updates I would not likely use any

such options. After all, do I really want to have the firmware reboot go
start vios, and vios start IBM i and then shut it all down again to

apply a

possible two stage update?

Of course not. But then, IBM delivers everything: Hard-, firm-, and
software. Is it so hard to set a flag somewhere so that even a two stage
update won't involve a full IPL of all subsequent LPARs?

My experiences come from the VMware ESXi world where I'm not aware of
multi-stage upgrades but a very comfy environment with automation being
normal instead of the need about coming up with more or less crappy
workarounds for things I'd expect to be already provided by the vendor.
Aside from that, there it's usually apply update, reboot host, done.

Anyway… the responses from you, Jim and Roberto reinforce my prejudice
about IBM Midrange LPAR handling being still inferior to VMware in terms

of

admin friendliness. Automation can be a measure to lessen human error,

when

applied cautiously and correctly. Why would anybody switch off/reboot a
server before LPARs/VMs have properly been shut down prior?

Now, I have my Power systems in remote data centers. Each a few hours

away. I can power cycle, update, etc all remotely and do so frequently.
I've done firmware, vios, IBM i, etc all remotely. The only time I go to
the DC is to move cables, install new hardware, etc. Doesn't really

happen

too often.

Same here, for the LPAR based system.

Basically the thought that having all the lpars correctly start/end in

the proper sequence in the unlikely event that these data centers with
multiple power inputs, multiple UPS, multiple generators, multiple

circuit

breakers is not worth the hassle when I can do it manually for the

minimal

chance of that occurring. And thus maintain control for the
firmware/vios/os upgrades.

Yes, the event is very unlikely. But with an increasing about of

machinery

to take care of I like to make the computer "think" for me. I'm human

and I

tend to forget things.

:wq! PoC

--
This is the Midrange Systems Technical Discussion (MIDRANGE-L) mailing

list

To post a message email: MIDRANGE-L@xxxxxxxxxxxxxxxxxx
To subscribe, unsubscribe, or change list options,
visit: https://lists.midrange.com/mailman/listinfo/midrange-l
or email: MIDRANGE-L-request@xxxxxxxxxxxxxxxxxx
Before posting, please take a moment to review the archives
at https://archive.midrange.com/midrange-l.

Please contact support@xxxxxxxxxxxxxxxxxxxx for any subscription related
questions.

--
This is the Midrange Systems Technical Discussion (MIDRANGE-L) mailing list
To post a message email: MIDRANGE-L@xxxxxxxxxxxxxxxxxx
To subscribe, unsubscribe, or change list options,
visit: https://lists.midrange.com/mailman/listinfo/midrange-l
or email: MIDRANGE-L-request@xxxxxxxxxxxxxxxxxx
Before posting, please take a moment to review the archives
at https://archive.midrange.com/midrange-l.

Please contact support@xxxxxxxxxxxxxxxxxxxx for any subscription related
questions.