Solaris

I have a single zpool with 3 2-way mirrors ( 3 x 2 way vdevs) it has a degraded disk in mirror-2, I know I can suffer a single drive failure, but looking at this how many drive failures can this suffer before it is no good? On the face of it, I thought that I could lose a further 2 drives in each pool, or even 5, but then I read that if any vdev completely fails in any zpool the entire pool is toast.

Code:

pool1 DEGRADED 
mirror-0 ONLINE
c0t5000C5004835382300 ONLINE
c0t5000CCA02533A897d0 ONLINE
mirror-1 ONLINE 
c0t5000CCA03CA35A25d0 ONLINE
c0t5000C5004867DF01d0 ONLINE
mirror-2 DEGRADED
c0t5000CCA06E11B840d0 ONLINE
c0t5000C5004821F08Fd0 UNAVAIL

Moderator's Comments:

Please use CODE tags as required by forum rules!

Your pool can still support two additional disk failures, assuming they are all parts of different mirrors. However, if you lose the second disk of mirror-2, or any pair of disks from the same mirror, your pool is lost and you need to recover from backup.

Thanks for the reply, I now understand.

If I lose another disk in vdev mirror-2, then it's gone.

I guess what I don't fully understand is how the read/write process works across the pool.

For example, a read/write process goes off to pool1, what determines how the data is written across the 3 vdevs I have? Does it just spread it across the 3 vdevs?

ZFS is striping data on the top three vdevs.

There is way to guess which one it will select for a given single record write but on average, writes will be balanced on all devices.

I thought as much, just didn't know for sure!

Thank you for sharing your knowledge.