RAID 6 and hot swap

[NdridCold]

If you have 2 drives fail in a RAID 6 setup with 6 physical drives can you hot swap them out at the same time or do you have to rebuild the array one drive at a time?

 

[ss202sl]

replace 1 at a time. Rebuilding will put stress on all the drives, and rebuilding 2 at the same time will add more stress and take longer(risking failure of a third drive). Replace 1 and let it rebuild, and then replace the other.

 

[NdridCold]

But theoretically it can be done?

Also have another question. Little different scenario. Still have a RAID 6 but only one failed drive with another showing signs of failure, so:

PD1= Failed Firmware State
PD2- Media errors but not in a failed firmware state

Question is would PD2, if hot swapped be rebuilt from parity even though its technically not a failed drive or if I want to replace that drive would I need to shut down the server, clone the drive and stick it back in? Or just wait until it completely goes belly up then swap it out?

 

[ss202sl]

Yes, it can be done.

If you have a drive that is acting up, replace it and let it rebuild. The whole thing with raid, is that it is high availability(of the data) - you shouldn't have to power down(and remove access to data) because of a single disk problem.

 

[NdridCold]

Ok. That answers my question. I was just not clear on if a drive needs to be in a certain state in order to trigger the parity rebuild if you put another (new) drive in. Thanks!

 

[TyrOd]

DO NOT attempt to rebuild a RAID 6 array with a drive that has media errors included.
It WILL cause inconsistent parity problems.
Assuming you have a backup, you have 2 safe options.

1. Remove both unstable drives and rebuild 1 at a time to 2 spares.
2. Remove both unstable drives, get a 100% sector by sector image from the drive with media errors on to a spare, then replace both drives to do a rebuild.

If you DON'T have a independent dedicated backup, then you have 1 safe option:

Remove all drives and make sector by sector images and then proceed to the above steps for rebuild.

 

[TyrOd]

Ok. That answers my question. I was just not clear on if a drive needs to be in a certain state in order to trigger the parity rebuild if you put another (new) drive in. Thanks!

Unless you remove both unstable/failed drives you have little assurance that the rebuild won't occur without outdated/corrupt parity data from the drive with media errors.

The safe option is to rebuild 1 at a time.

 

[NdridCold]

1. Remove both unstable drives and rebuild 1 at a time to 2 spares.

Thanks Tyr0d - what do you mean "rebuild 1 at a time to 2 spares"