Theoretical loss of two hdds in raid z2

[Bhoot]

I am sorry if it has been discussed before but I couldn't find it in threads that I went through.
How I came to think of this was during my recent hard disk replacement. One of the WD reds got 647 unreadable (pending) sectors within the last 2 months. Now a new HDD has been slapped in and the 'broken' disk has gone for RMA. This brought me to the next conclusion that right now, while the disk is being resilvered, my system at its weakest. Even with the disk with 647 errors the data still existed on that disk. Right now the new disk just has over 50% resilvered data.
Now theoretically if I were to lose a different (old) hard disk would I have any data loss? Theoretically there should be none in a raid z2 since it allows for 2 disk parity but I wanted to hear from the people who have suffered from such failures or those who have simulated such failures on test benches. Also please separate bit rot from data loss in the provided answers.
Also want to know if a raid z2 shows 2 disks resilvering during a scrub am I under any threat of losing data or having bit rot?

 

[m0nkey_]

First things first. Backups. You have them right? Having an array of drives does not count.

Secondly, with RAIDZ2 you can lose two drives before your pool is considered degraded. By losing one drive, the pool effectively becomes RAIDZ1. Loosing two, that's just bad luck (hence the backups). As for re-silvering during a scrub, it's not the best idea in the world. You should really wait until the re-silver completes.

For bit rot, if you're meeting the hardware recommendations (recommended server board, Intel, ECC, etc.), you have nothing to worry about.

 

[Bhoot]

First things first. Backups. You have them right? Having an array of drives does not count.

Yes I have kept the really IMPORTANT stuff on a WDMBLD on a raid1 so data is safe :)

Secondly, with RAIDZ2 you can lose two drives before your pool is considered degraded. By losing one drive, the pool effectively becomes RAIDZ1. Loosing two, that's just bad luck (hence the backups). As for re-silvering during a scrub, it's not the best idea in the world. You should really wait until the re-silver completes.

The thing is that my scrub schedule is coming soon (threshold). Will the scrub start if the resilvering is on or should I change the schedule?

For bit rot, if you're meeting the hardware recommendations (recommended server board, Intel, ECC, etc.), you have nothing to worry about.

Do have intel, ECC, WD Reds and asus mobo [all server grade]

 

[danb35]

with RAIDZ2 you can lose two drives before your pool is considered degraded.

This is not correct--ZFS will mark the pool as DEGRADED when a single disk is offline or removed. Your pool doesn't need to have lost all redundancy to be degraded, it just needs to have less redundancy than designed.

 

[jgreco]

Also want to know if a raid z2 shows 2 disks resilvering during a scrub am I under any threat of losing data or having bit rot?

You're only under threat of losing data if ZFS is not able to retrieve your data from *somewhere*. Two disks being totally removed from a RAIDZ2 eliminates redundancy, and ANY read error on the remaining disks could result in an uncorrectable error. That's why for a larger array, we use RAIDZ3 ... because experience says failures like to crop up in small batches.