Is there any place for enhancement requests/ Enhancements to the shutdown command

[Wolf-P]

It is quite quirky to get an UPS to shut the MSA down in the right moment.

I have an expect script for shutting down the MSA, which works nicely. Problem is, that this script has to run on the appliance/computer/device, which manages the shutdown process and should be run at the end of (or better after) the shutdown of all hosts/computers connected to the UPS, which triggered the shutdown. This is easy, if the software runs on a device not connected to the MSA, but that is overkill in smaller setups and if the host hosting the shutdown app/vm shuts down you cannot shutdown the MSA afterwards. So you need to test the shutdown sequences and stop shutdown times several times to find an acceptable - but not optimal - solution.

This task could be very easily accomplished in a perfect way, if the shutdown command in the SSH shell had the possibility to set a parameter to delay the shutdown for a given number of seconds - or better a switch to wait with shutdown until all host connections are lost/down. And as an additional feature an option like a 'force' parameter to bypass the question, if you really want to shutdown the controllers.

Just my thoughts on this issue.

[Shawn_K]

Hello,

I am unsure of what you are exactly asking for. The MSA has a shutdown command that will work using a script.

Yet it sounds like you want a parameter on the MSA to detect when the hosts have been shutdown. That is rather an impossible task. The MSA can detect when a host is no longer present, but that could be caused by an issue on the host/server/VM. The host might have been shutdown, deleted, removed, migrated, or gone offline due to external issues. The MSA cannot detect all of those situations and differentiate between them. I don't know of any arrays that have that ability.

Any Enhancement requests should be filed by your Account Team or Partner Vendor. They will have the ability to fill out all the required forms and submit to the appropriate Program Team to review that request. Once reviewed a decision will be made whether the request is possible or not.

I work for Hewlett Packard Enterprise. The comments in this post are my own and do not represent an official reply from HPE. No warranty or guarantees of any kind are expressed in my reply.

Cheers,
Shawn

[Torsten.]

If all servers are down there is no longer data in cache.

And even if there is data in cache, the MSA will save the cache to flash.

There is no risk to loose data, even if you don't do a shutdown (in case of emergency, of course).

[Wolf-P]

Thank you for your answer!

No, I don't need such a sophisticated solution. It is sufficient to test if any host is present. If there is no host present, regardless of the reasons for that situation, it is safe to shutdown the storage controllers - and that's all I want. Even in a planned shutdown, this could be helpful, if you forgot to shutdown a connected host or accedentally restartet one host instead of shutting it down.

But the easiest thing would be just a timer, to allow a delayed shutdown. That would eliminate the problems of scripts running on other systems being:

- the other system has to remain up during the whole shutdown procedure,

- all switches in between have to work without interruption

- the ssh-session on the MSA has to remain open and is not closed because of inactivity

- (I did not test it via ssh, but you have quite difficulties logging into the web-interface, if no dns server is reachable - and of course I used the IP-address and not the dns-name in the URI)

That's just my thoughts on this topic, but the way via the Partner Vendor is too complicated for me.

[Torsten.]

I seriously doubt if you can be sure there is no host connected via FC, iSCSI or SAS.

[Wolf-P]

For SAS I don't know, how and if it is possible to check the link status on the MSA side. But for iSCSI and FC connections this is possible. And the MSA seemingly notices this, as it sends me everytime a notice, that there is no active host connection, if I shutdown one of the connected hosts.

Regarding powering off, without shutting down the controllers. I presume, that your statements are correct, as that's the idea of the battery/capacitor backed cache. But the documentation tells another story. It probably depends on how long it will take to regain power. But there seems to be a compact flash memory in each controller, to which data in the cache is saved in case of a power failure. So in theory the cache data should survive a long time of power outage.

[Shawn_K]

Hello,

You always want to issue a shutdown command on the MSA to allow any data to flush to disks. It is dangerous and can cause data loss, to pull power on the array The MSA does have battery backed cache to help preserve cache data in case of a sudden power loss. This does not mean the battery back cache lasts indefinitely. Restoring power and allowing the cache to flush should be done as soon as possible.

The shutdown command for the MSA is not configurable with a variable to allow you to delay shutdown. Even our Enterprise arrays do not have that option.

My suggestion would be to run a script off a server/VM that is not being shutdown. Additionally, unless you are performing maintenance, I do not know why you would want to shutdown your entire SAN at the same time.  This type of use case is limited, which is likely why this feature is not available for most storage arrays.

I work for Hewlett Packard Enterprise. The comments in this post are my own and do not represent an official reply from HPE. No warranty or guarantees of any kind are expressed in my reply.

Cheers,
Shawn

[ArunKKR]

Hi,

As mentioned by Shawn, Any Enhancement requests should be filed by your Account Team or Partner Vendor.

None of the MSAs (G1 to G6) has this feature.