RAID5 failure: 2 bad HDD's at the same time

[CapLeaker]

Re: RAID5 failure: 2 bad HDD's at the same time

No, not all of my contents are lost. I was able to borrow another similar NAS. Over Prime Day on Amazon I've ordered 4 new HDD's. Plan is to "replace" the old raid with a new raid. Basically the old raid goes into the borrowed NAS. My NAS is getting all new drives then copy everything back.

[Curious.George]

Re: RAID5 failure: 2 bad HDD's at the same time

No, not all of my contents are lost. I was able to borrow another similar NAS.

Perhaps you see why I have TWO of every piece of hardware?

Over Prime Day on Amazon I've ordered 4 new HDD's. Plan is to "replace" the old raid with a new raid. Basically the old raid goes into the borrowed NAS. My NAS is getting all new drives then copy everything back.

If you feel ambitious/inquisitive, you may want to keep the RAID drives (after you've returned the borrowed equipment hosting them, presently) and "play" with some of the RAID recovery tools that are available. Your data will already have been recovered so the drives' contents are "disposable" (?). I.e., you can play with them (and software recovery tools) without risk of LOSING anything.

That could give you a head start if you find yourself in a similar situation at a future date (i.e., you will KNOW that it can be done and HOW to do it -- instead of HOPING it can be done and stressing over HOW to do it!)

Interesting experience, eh?

[CapLeaker]

Re: RAID5 failure: 2 bad HDD's at the same time

I am building a new RAID now. Coping files like stupid! Well I am not sure if Iam going to play with the old RAID drives, since they are still in warranty and I like to get them replaced if I can. I didn't go for a new NAS. I just went for new HDD's. Maybe I look for a new NAS later in the fall.

[Curious.George]

Re: RAID5 failure: 2 bad HDD's at the same time

I am building a new RAID now. Coping files like stupid! Well I am not sure if Iam going to play with the old RAID drives, since they are still in warranty and I like to get them replaced if I can. I didn't go for a new NAS. I just went for new HDD's. Maybe I look for a new NAS later in the fall.

You might want to consider setting up the drives as JBOD and using some number of them "offline" to hold backups.

RAID is good for enhancing throughput (RAID0) or enhancing availability -- sort of like ECC RAM enhances the availability of data stored in "memory" (but, you wouldn't RELY on ECC RAM for long term storage/backup).

If you can tolerate the NAS (as JBOD) throwing an error from time to time and using that to prompt you to drag out the "backup", this may be a more effective use of the platters. Your AVAILABILITY goes down (cuz the data wasn't reliably present when you wanted it) but your DURABILITY goes up (cuz the data wasn't lost!).

I'd still advocate playing with some of the recovery tools just to see if you'll have a fallback "out" in the future... (some vendors just roll a Linux distro into their appliances; others reinvent the wheel -- or, deliberately obfuscate some standard implementation just to tie you more closely to them)

[CapLeaker]

Re: RAID5 failure: 2 bad HDD's at the same time

This is the idea I have for later. I am going to upgrade sooner or later my NAS. So I going from a 4 bay to an 8 bay NAS. Run a 6 drive RAID 6 and with the other 2 bays two large capacity HDDs on JBOD.

[Curious.George]

Re: RAID5 failure: 2 bad HDD's at the same time

This is the idea I have for later. I am going to upgrade sooner or later my NAS. So I going from a 4 bay to an 8 bay NAS. Run a 6 drive RAID 6 and with the other 2 bays two large capacity HDDs on JBOD.

I'd still question the appropriateness of ANY RAID. Remember, all it provides is AVAILABILITY, not DURABILITY. The extra drives might be better used in another configuration (or even "kept cold")

I've set up RAID on my SAN (and only on my SAN) because it essentially provides "C:" for the many virtual machines that are stored on it. When I activate a VM, I don't want to risk the VM hiccuping because of a disk error (e.g., a corrupt binary or otherwise unreadable file). But, I don't rely on the RAID to preserve my data any longer than that particular "session" (if the array crashes just as I log off, I don't care -- I'll see which drive(s) have failed, replace them and then reinitialize their content from my archives).

[Stefan Payne]

Re: RAID5 failure: 2 bad HDD's at the same time

RAID is good for enhancing throughput (RAID0) or enhancing availability -- sort of like ECC RAM enhances the availability of data stored in "memory" (but, you wouldn't RELY on ECC RAM for long term storage/backup).

Well, it also, in theory, should help against one drive failures.
The Problem is that if one drive fails, another one shortly follows...

And you have to know the advantages and disadvantages of RAID and the RAID Levels. And that if you have a 4 Drive RAID5 with 4 drives of the same type, the wear might be similar on all of them, so they might fail in short succession.

But RAID is _NEVER_ a backup. And must not be seen as such...
Well, RAID1 maybe, kinda...

[eccerr0r]

Re: RAID5 failure: 2 bad HDD's at the same time

No, RAID1 is not backup, not any more RAID3, RAID5, or even RAID6.

An off disk backup will protect you from rm -rf / . RAID of any level will not protect you.

There is something to be said with versioning filesystems as a type of backup. Coupled onto some sort of redundant RAID (i.e., not RAID0), this however does classify as a weak backup system. Still does not protect you from metadata loss of the versioning filesystem.

I've been contemplating a versioning filesystem, but still will need to snapshot them to another disk set. Currently I only snapshot my RAID5 to another RAID (RAID1, incidentally) -- up until the capacity of the RAID1 gets exceeded...then it's time for a new disk...or a file purge...

[Curious.George]

Re: RAID5 failure: 2 bad HDD's at the same time

No, RAID1 is not backup, not any more RAID3, RAID5, or even RAID6.

An off disk backup will protect you from rm -rf / . RAID of any level will not protect you.

ANY "backup" protects you (there's a reason tape is still used!).

And, the example suggests the user may have CARELESSLY typed "rm -rf". Had he intentionally (or unintentionally/carelessly) done any number of OTHER things he could still lose data.

Change one line of code in a program -- then, try to remember WHICH line it was and what the original version happened to be!

Crop that photo of your kid's birthday party -- and then try to recreate the imagery that was lost!

Boost the gain in an audio file (so parts peak above 0dB) and then try to recover the original signal. Or, attenuate it and try to re-boost it, later, without dragging the noise floor up in the process.

RAID (except 0) gives you availability -- it makes the data stored more accessible in a variety of adverse conditions. Just like ECC helps make RAM data available in adverse conditions.

[Stefan Payne]

Re: RAID5 failure: 2 bad HDD's at the same time

RAID protects against drive failures, kinda...
But you need to know the disadvantages or the effect it has on the other drives!

For example, if you have a 4 drive RAID5 and one drive fails, you can still work.
But if you have 4 identical drives, chances are the other 3 are also on their last legs and will die soon as well. As the Wear and Tear on all drives is similar...

So you need to update your backup ASAP, get 4 new drives (or if its older, you might be able to get 2 and go for RAID1 or 3 Drive RAID5 or 4 Drive RAID6 instead) and copy the content of the RAID to the new storage.

The old RAID you either keep for safety/emergency Backup or you destroy the drives. But you must not rebuild the array if its older than a couple of weeks...

if you make it new and after a week, one drive fails, you can think of replacing the disc and rebuilding the Array...

[CapLeaker]

Re: RAID5 failure: 2 bad HDD's at the same time

Right now I have 2 RAIDs. A Raid 5 and a RAID 10. We will see how that goes.

[Curious.George]

Re: RAID5 failure: 2 bad HDD's at the same time

Right now I have 2 RAIDs. A Raid 5 and a RAID 10. We will see how that goes.

Nothing "wrong" with using RAID. My SAN has four 6-drive RAID6's in it. As I'm primarily using it in a R/O application, the "write penalty" doesn't come into play.

The real issue is NOT relying on it to provide your (sole) backup!