Failed HDD, now working, do I trust it?

[CHR1S]

Ive had a server HDD fail in raid 5.

I removed it, cleaned and checked it and re-installed. Its now happily rebuilding and seems ok.

There are no recorded errors (Media and Pre Fail)

 

Do I just leave it and see how it goes?

[sted]

ive never had one fail and be reusable for long in my experience if it lasts much past rebuilding youre doing well and in 1 case recently got to 90%ish rebuilt and took the server down until it was pulled out

[CHR1S]

I will keep everything crossed then! haha

[pcstru]

Ive had a server HDD fail in raid 5.

I removed it, cleaned and checked it and re-installed. Its now happily rebuilding and seems ok.

There are no recorded errors (Media and Pre Fail)

 

Do I just leave it and see how it goes?

 

I'd take a view dependent on what it is being used for. We have some NAS type devices used as part of the backup (prior to tape), if it was one of them I'd probably be happy to take the risk. On our main SAN, degrading the array has a performance impact on rebuild and I'd want to avoid that. Worse, a disc that is dodgy but not being failed by the RAID controller (i.e. subject to retries) can have an ongoing impact on performance - again, I'd want to avoid that on our primary storage.

[apeo]

To be fair its 50/50 for me on how successful it rebuilds or not but those that did rebuild lasted quite a while. Guess you have to weigh how critical the data and the server is, then decide if its worth the potential risk that this disk will potentially fail in tandem with another.

[CHR1S]

Its the servers main storage, but, its backed up twice and in a DFS.

[elsiegee40]

Given that it's part of a RAID array, it's failure shouldn't be a disaster as, in theory, the server will keep going.

 

It is, however, a warning. It could well die permanently and the other drives could well be on the verge of failure too... or they could march on for years... there is absolutely no way of telling.

 

Losing one drive with RAID is not a problem, losing two potentially is. Personally I'd replace the drive and keep this one as an emergency spare to tide you through should you have a permanent failure on another drive.

[apeo]

Its the servers main storage, but, its backed up twice and in a DFS.

 

When you say its a DFS, is it replicated? Given you have backups and if it is replicated then guess it isn't an issue. Personally I'd go the same route as elsiegee40, and replace the drive if I could, using the potential failing one as a spare. I'd do that just so I didnt have the hastle of rebuilding a server.

[CHR1S]

Yes its a replicated DFS, im erring on the side of caution now too!

[CHR1S]

Rebuilt and passed its tests with no issues. Will watch it like a hawk! haha

[pete]

I wouldn't have done - I'd have let it rebuild on the spare and phone $vendor for a replacement disk. First-line may quibble about testing the disk...second-line won't.

[gshaw]

I'd swap that out, especially if it's under warranty... better safe than sorry imo

[jinnantonnixx]

It's let you down once already. Why do you think it won't let you down again? You've had your warning!

[FragglePete]

Usually the first thing HP Support get you to do if you have an orange light flag up on your drive cage; remove from cage and plug back in again. If it stays orange then they try and get you to run further tests, but at that point I usually stamp my feet and demand a new hard drive. Watch it; like a hawk.

 

Pete

[twin--turbo]

Done pull - push - rebuilds many times and disks have either not rebuilt, died soon after or lasted for years.

 

It all depends on why it droped out of the array in the first place.

 

Have a spare on the shelf or as a assigned hot spare.

 

 

Rob

[rad]

And put a note on the drive itself with a date, incase it goes again and you cant remember if it was 5 or 4, at least you know to swap the drive over.

[twin--turbo]

And put a note on the drive itself with a date, incase it goes again and you cant remember if it was 5 or 4, at least you know to swap the drive over.

 

That's on the helpdesk

 

Rob