topic Re: MSA2062 Test Controller and HDD Failure in MSA Storage

MSA2062 Test Controller and HDD Failure

sondavid — Wed, 13 Oct 2021 05:56:53 GMT

Hello community,

we would like to simulate the failure of controller and hard drive on HPE MSA2062

My question, how can we do that? has someone already done? jus Bestpractice.

I just want to pull out a hard drive, see if everything works, and plug the hard drive back in.

Unplug the controller and see if everything works and plug it in again.

It's OK?

Thanks

Re: MSA2062 Test Controller and HDD Failure

JonPaul — Tue, 12 Oct 2021 16:57:13 GMT

@sondavid
You can do just as you suggested and physically pull the drives and/or controllers. Both are HOT-SWAP components. When doing this, bear in mind you should allow them to be out for a little while as the system has some timeouts to step through when components disappear and their partner's to take over.
I would caution you against doing degradation testing with a live, production system and data as you could also have a real fault in the middle of the test which could take you offline. Ex you pull a drive and in the time the drive is removed another drive faults and all data could be lost.

Re: MSA2062 Test Controller and HDD Failure

sondavid — Mon, 18 Oct 2021 09:50:09 GMT

@JonPaul

Thanks for the answer. I will pay attention to that.

Another question:

we want to connect another esxi host to MSA. Do we have to pay attention to something here?

I think no right? please correct me if i'm wrong.

Thanks

Re: MSA2062 Test Controller and HDD Failure

JonPaul — Mon, 18 Oct 2021 14:26:19 GMT

Should be no worries connecting another host (ESX or other). The MSA 2060 systems only allow explict mapping/attaching of LUNs which will limit the visibility of one host's LUNs from any other host.
Just so you are aware there is a Best Practice check available for the MSA storage systems for the last 3 generation systems available here: https://www.hpe.com/storage/msahealthcheck
A couple of the checks are surrounding mapping, making sure that HOSTs are mapped redundantly and that LUNs are not mapped default (meaning that any initiator connected would have access, a possibility in older MSA generations)