After power loss, the RAID config changed, some drives marked as foreign, but all data still there

configurationdegradeddell-percraid

I have a Perc 5/i in a Dell Precision 590. The controller DOES have a battery backup, and the controller reports that its status is healthy. There are five 1TB SATA drives connected to the controller. My original config looked like this:

  • VD0: RAID5
    • PD0
    • PD1
    • PD2
  • VD1: RAID1
    • PD5
    • PD7

Fairly simple setup. Last week we had a power failure and I was notified that both VDs were degraded. So I investigated. The configuration has changed. Here is the new configuration:

  • VD0: RAID5
    • PD0
    • PD1
    • PD5
  • VD1: RAID5
    • PD0
    • PD1
    • PD5

This makes no sense. How is this even possible!?

Note: The PD Mgmt screen says that Discs 0, 1, and 5 are online and 2 and 7 are "foreign" (what does that even mean???).

Ever weirder, much to my surprise: The VDs both seem to be completely functional and all of my data is intact and my system runs fine. I'm at a loss for words.

I have the data backed up, so it won't be The Worst Day Ever if I lose it and have to restore. But I'd really rather not have that happen. If this had been a normal drive failure, I'd know what to do. But this state is so fubarred I'm afraid to even touch it. Help?

Here are some "screenshots" (for lack of a better word) that hopefully will add some context / additional information if needed.

RAID config for VD0
RAID config for VD1
PD Mgmt
Foreign View

Best Answer

Although the config shown in the PERC BIOS looks all wrong, it's clearly still able to properly use the right configuration because you have access to data. The odd data displayed from the BIOS is likely just due to a firmware bug caused by some damage that happened to the configuration data on some of the disks.

You should never import foreign config on a PERC controller if you can see all data from the OS (no missing/offline VDs), no matter how strange things look from the PERC BIOS.

Foreign config is usually seen in two scenarios:

  • The configuration data was somehow damaged on one or more drives, and the controller recognizes a mismatch with the copy of config data that it keeps cached.
  • The controller booted and found that its cached configuration was completely missing or damaged, and had to refer to the disks to detect the config (this usually results in all drives showing as foreign)

An import essentially takes the config data on a "foreign" drive and uses that to replace the config data on the other drives it claims to be part of a VD with. In some scenarios, this can leave you with an offline VD instead of just a degraded one.

I would recommend that you ensure your PERC's firmware is up to date. There are no steps you can take to resolve this which have zero risk of causing you to have to restore from backup at this point though, so you might also consider getting those backups up-to-date...

Related Topic