How much time do I have to replace a failing hard drive in RAIDz?

[Wade]

I started my Freenas box with 4 2tb hard drives, setup as RAIDz2. Today I am adding another 2tb drive, rebuilding the system new and had intentions on staying with RAIDz2. However, the ZFS volume manager is suggesting that I go RAIDz. From what I've read, RAIDz is faster and offers more storage. When I select RAIDz2 it says that its "not" optimal. From my research it seems more users prefer Z2 over Z (at-least those who were victims of drive failure). My question is, how much time do I have to replace a failing hard drive in RAIDz? and also, is there a way to make my freenas box shutdown on that type of error? giving me more time to replace the failing drive, sometimes I could be away from the server from 12 to 24 hours. Its just a home server and doesn't necessarily need to be on 24/7

Any advice on raid selection with 5 2tb drives is appreciated.

 

[fracai]

The "non-optimal" statement about five drives in Z2 can probably be ignored. The performance difference for a home user probably isn't worth worrying about. If you do care about it, either go Z1 or get another two TB drive as six drives in RAIDZ2 is an optimal configuration.

As for how much time you have, it's basically however long it is until your next drive fails. If you have a Z1 pool and one drive fails, you'd ideally start resilver with a new drive immediately. If a second drive fails before the resilver completes, you'll have lost your pool. If you run the system in a degraded state for a year before you finally replace the failed drive and the resilver completes without error you're safe.

RAIDZ2 is preferred because it allows a second drive to fail during the resilver without losing any data. RAIDZ3 would allow three drives to be down at a time. It's all a matter of how likely the hardware is to fail.

I'm not aware of a process that would shut down the system when a drive fails, but I'm sure you could write up a script that would probe "zpool status" and initiate a shutdown if the pool enters a degraded state.

 

[joelmusicman]

RAIDZ1 is DANGEROUS in a drive failure scenario. If you get even 1 or 2 bit errors (which is not at all unlikely) while resilvering you can lose EVERYTHING.