Best way to handle drive in array "predicted to fail soon" warning on Exchange server?

I have a three disk RAID-5 array on an older E200i HP card. One of the disks is showing the "drive is array predicted to fail soon" error in ACU and has the blinking amber light. I've ordered a new hard drive that will arrive today, but I'm trying to figure my best option.

1. Should I pull out the predicted to fail drive and replace it with this one and, I assume, it'll automatically start to rebuild the array?
2. Should I just add the new drive as a hot spare in an empty slot and just wait for this other drive to potentially fail?
3. ???

Is a 'predicted to fail' drive likely to have an impact on hard drive performance? Our Exchange server has been much slower the past week, so I'm assuming so, but we do have other problems with this server (it's 8+ years old), so I didn't know if this is still the most likely cause of the recent latency issues. If it is the case, I would likely want to do #1 after a full backup, but at the same time, it's not actually failed yet, so I'm a bit undecided.
ruhkusAsked:
Who is Participating?

[Webinar] Streamline your web hosting managementRegister Today

x
 
Handy HolderConnect With a Mentor Saggar maker's bottom knockerCommented:
Backup just for safety sake and then swap it. Check the part number / firmware version first though since there are some versions that report SMART errors when there's nothing wrong with the drive. Also look through an ADU report for bad sectors on the other disks that would stop a rebuild.

Setting as a hot spare is not a good idea since your controller treats hot spares as temporary so that would involve rebuilding parity twice.
0
 
frankhelkCommented:
It would probably save you some money if you install the new drive as hot spare and let the RAID controller decide what to do - you'll press the rest of life out of the old drive .... If you're unlucky, it would just kick out the old one and activate the hot spare as soon as the hot spare is available :-). Anyhow it's better than waiting for a failure.

On the other side - HDDs are not that expensive at all, so if you value the uptime of your system, just buy two disks, replace the weak drive and add the other as hot spare. That would be the most safe option.
0
 
frankhelkCommented:
Addendum: The failing drive might impact the system performance of the station. Depending on the type of problem, the controller needs to do additional actions to compensate the errors (re-read, extract data from another drive, relocate bad sectors, the need to read sectors more than once, etc.) - that might slow down the RAID considerably.
0
 
rindiCommented:
First make sure your backups are OK, then just pull out the failing disk and insert the new one (If your servers disks are hot-swappable, don't turn off the server). Normally it should just rebuild to the new disk. If it doesn't, assign it to be a hot-spare and the rebuild should start.

Make sure you do this after hours, as a RAID rebuild causes the system to slow down which the users may notice.
0
All Courses

From novice to tech pro — start learning today.