Hpe 3par 8400 OS upgrade from 3.2.2mu6 to 3.3.1 mu3

[dockerd]

Code: Select all

Hello, if you are live, then you should upgrade from system patched with all available patches. If you are not live, then you can upgrade directly.

good
i'm live, you means patches that found in hpe upgrade recommendation document and marked as mandatory patches ??

Code: Select all

I have experience upgrading from 3.2.2 MU4 with only some patches installed directly to 3.3.1 MU3 without any problems.

Can you share this experience , please

Thank you

[ailean]

Code: Select all

P patches don't need node restarts and can we installed one after the other okay, some have dependencies so the order installed can matter and some require running commands at the CLI after to finish and check they completed (the release notes will say so if required).

Ok.
Do patch update need any pre-requisite except patch dependencies such cpu and memory utilization less than 50% , stop all task , stop all activity and disconnect any ssmc sessions ???

Code: Select all

Version changes (e.g. 3.2.2 -> 3.3.1) or MU installs (e.g. 3.3.1MU1 -> 3.3.1MU2) do require node reboots. One node at a time and recommended to use the pause option so you have time to check hosts have reconnected to the rebooted node okay before doing the next in an online live environment.

Ok.
What you mean by pause options ???

Thank you so much

Typically you don't need to worry about array activity/state when doing just Patches, it will do a health check first so as long as nothing broken can proceed. You may experience SSH/SSMC sessions dropping out if those services are restarted by the patch, normally SSMC will recover automatically. It's worth letting any admins know about patching so they don't think something broke.

All the things you mention are more relevant to MU and version upgrades because of the node reboots.

The pause options are for during node reboots, without selecting the pause option the 2nd node will start updating/rebooting as soon as the 1st node rejoins the cluster (after its reboot). With the pause option you have time to check the paths are restored okay on the hosts and do manual rescans on any hosts that failed to notice the 1st node return. Depends how critical the hosts are, we've tried to keep all hosts updated and configured within the 3PAR recommendations and other then one small VMware cluster have never had an issue during updates in the last 10 years or so.

[dockerd]

Code: Select all

  Typically you don't need to worry about array activity/state when doing just Patches, it will do a health check first so as long as nothing broken can proceed. You may experience SSH/SSMC sessions dropping out if those services are restarted by the patch, normally SSMC will recover automatically. It's worth letting any admins know about patching so they don't think something broke. :)

All the things you mention are more relevant to MU and version upgrades because of the node reboots. 

Ok

Code: Select all

.  The pause options are for during node reboots, without selecting the pause option the 2nd node will start updating/rebooting as soon as the 1st node rejoins the cluster (after its reboot). With the pause option you have time to check the paths are restored okay on the hosts and do manual rescans on any hosts that failed to notice the 1st node return. Depends how critical the hosts are, we've tried to keep all hosts updated and configured within the 3PAR recommendations and other then one small VMware cluster have never had an issue during updates in the last 10 years or so.  

How i can activate the pause options ???

Thank you

[ailean]

Code: Select all

  Typically you don't need to worry about array activity/state when doing just Patches, it will do a health check first so as long as nothing broken can proceed. You may experience SSH/SSMC sessions dropping out if those services are restarted by the patch, normally SSMC will recover automatically. It's worth letting any admins know about patching so they don't think something broke. :)

All the things you mention are more relevant to MU and version upgrades because of the node reboots. 

Ok

Code: Select all

.  The pause options are for during node reboots, without selecting the pause option the 2nd node will start updating/rebooting as soon as the 1st node rejoins the cluster (after its reboot). With the pause option you have time to check the paths are restored okay on the hosts and do manual rescans on any hosts that failed to notice the 1st node return. Depends how critical the hosts are, we've tried to keep all hosts updated and configured within the 3PAR recommendations and other then one small VMware cluster have never had an issue during updates in the last 10 years or so.  

How i can activate the pause options ???

Thank you

Normally we have HPE do all our MU/Version upgrades, from what I remember when you go to start those types of installs from the SP you see I think three options, Offline Install (does all nodes at once and restarts array), Online Install (does one node after another) and Online Install with pauses (does one node, waits, does another node).

Not sure of the wording on the screen but each option does have a description and how the options are presented (tick boxes/selection boxes) varies a bit between SP versions.

[cali]

Code: Select all

Hello, if you are live, then you should upgrade from system patched with all available patches. If you are not live, then you can upgrade directly.

good
i'm live, you means patches that found in hpe upgrade recommendation document and marked as mandatory patches ??

Code: Select all

I have experience upgrading from 3.2.2 MU4 with only some patches installed directly to 3.3.1 MU3 without any problems.

Can you share this experience , please

Thank you

One example is the P149

HPE 3PAR OS 3.2.2 MU6 P149 addresses the following issues:
A SAS port stops responding and a controller node restarts while upgrading from HPE 3PAR OS 3.2.2 MU6 to 3.3.1 MU2.

[dockerd]

Thank you so much
I have 4 controllers with 3 disk enclosures and I'm planing to remove the last admitted enclosure before start the patching so advice how could we do that ?? Directly, unplug sas cables or i need do some from controllers them self