HPE EVA Storage

EVA8100 - Disk Groups and Disk Replacement

 
Steve Goodger
Occasional Contributor

EVA8100 - Disk Groups and Disk Replacement

We've recently been made aware of an issue from an HP rep in Australia.

We were informed that we should not replace disks in our EVA for upgrading capacity (e.g. slowly replace all 146GB FC with 300GB FC)

We were told that if we replace and insert a disk, that while the EVA is levelling should ANOTHER disk fail - we will lose the entire disk group.

Can someone please confirm? How does this differ from a standard disk fail and replace of the same size, and are we exposed to the same potential data loss in that scenario too?

That's a RAID-V failure scenario, but on an EVA?

Our belief is that there are far more redundancies built in to protect against this sort of failure.

Would seem to me that it would be counter-productive to have a disk group of too many disks since you then increase your chances of failure.

Thanks
Steve

2 REPLIES 2
Marcus Schack
Trusted Contributor

Re: EVA8100 - Disk Groups and Disk Replacement

Hello Steve... This sounds a little off to me. As long as you have your "sparing" level set and you are not removing too many disks at once, which might cause you to run out of space. You should be fine. I do this all the time.

So if you want to be really careful, set your sparing level to 2. And remove your disks a couple at a time. Then when you get a group of eight. Add all eight in on one go. You space calculations may be off a bit, but once everything is to one disk size, you will be fine. Also, make sure you have no Vraid0 LUNs. Because those will fail.

Remember, slowly does it. Don't over do it, it is not a job to be rushed.
Víctor Cespón
Honored Contributor

Re: EVA8100 - Disk Groups and Disk Replacement

Many people confuses rebuilding and leveling.

After a disk failure, controllers perform a RAID rebuild, regenerating all the data using the other copy or the parity. After the rebuild finishes, you are back on a fully redundant state (two copies of data for a RAID 1, for excample). If another disk fails after rebuild is finished, no data will be lost and another rebuild will be triggered.

Leveling moves data from one disk to others, but does not change the data, it has no relation to the data redundancy.

Also, you can ungroup the 146 GB disks before removing it, so they will not contain any data when removed.