RAIDZ: x3 3TB, 1 drive bad?

[fanix]

Hi,

I have a RAIDZ redundancy configurations, x3 3TB drives, which gives me around 6TB of data.
SMART has been detecting bad sectors on one of the drives for some time now. I want to RMA the drive, but I'm not sure if my zpool will function if I take the damaged drive out? I know it will show as degraded, but will I still be able to use the pool?

Also, is there anything I need to do before I take the drive out?

After I receive the replacement drive, I know to connect it to the exact same sata port then issue the following command: (replacing tank and clt1d0 params)

Code:

zpool replace tank c1t1d0

 

[cyberjock]

No, you should not be doing a replacement with that command. Please consult the FreeNAS manual for the proper way to replace a disk in FreeNAS.

As for your pool, there is a good chance you will be unable to rebuild your pool without having a problem. If your data is important you should make it a priority to backup your data now. With RAIDZ1 you should have a replacement disk ready before taking the current disk offline because you will have no redundancy until the resilvering process completes with the new disk. This will be a dangerous time for your pool. Many people do not recover from this(which is why I have my warning in my signature which people still don't read). Your pool will continue to function thoughout this process unless you have any kind of URE or a drive actually fails. A failed drive may be a death blow for your pool. A URE could result in anything from a corrupted file to a loss of the pool.

I would buy a replacement disk, then do the RMA for the bad disk you currently have. When the new disk comes in you should test it and keep it safe on a shelf or somewhere where it won't be bothered in case you need to do another replacement later.

If your pool ends up damaged and you lose the pool you will already have 4 drives so you can also take the opportunity to create a proper 4 drive RAIDZ2.

Good luck!

 

[ZFS Noob]

What cyberjock said. Read the RAID5/RAIDZ1 is Dead link in his signature for the overview of the problem. There's another discussion underway here where a developer indicated that he tested FreeNAS's resilvering to test this, where:

• He created a RAIDZ1 pool.
• He removed a drive.
• He intentionally damaged a file
• He resilvered
• The process completed without loss of the pool, however data was still lost. Luckily ZFS was smart enough to tell him which file was damaged.

What I'm trying to say is this: RAID5/RAIDZ1 sucks with large drive capacities, and will result in data loss when resilvering a high percentage of the time, but your choice of ZFS might save you, kind of.
What I'd do is this:

• Back up everything on your pool twice, to two separate storage devices. (Why? Because data is important, and I'm paranoid, and I've done this long enough to see low-probability things happen with surprising regularity.)
• Buy two new drives
• Destroy the pool, remove the bad drive and RMA it, then rebuild the pool either as RAID10 or RAIDZ2.
• Recover your data from backup.

But I'm paranoid. And I've got H1N1, so I'm not thinking all that clearly.

 

[cyberjock]

But I'm paranoid. And I've got H1N1, so I'm not thinking all that clearly.

Should have gotten a good antivirus... ba-dum-tss!

Thank you, thank you! I'll be here all week!

 

[fanix]

THANKS for all the info.

I'll RMA the drive and buy another drive and then create a new RAIDZ2 pool :)

I've already backup up my current pool to another pool, so will just copy everything back once I get the two new drives.

QUICK QUESTION: If I just replace the one damage drive and keep the RAIDZ pool I currently have (x3 3TB), how long will the resilvering process take. I have about 3.5TB of data used in my pool. Capacity is 5.5TB

 

[cyberjock]

That depends on various factors. You'll have to find out when you do it. :P

 

[fanix]

SO it turned out to be way easier than what I thought. I simply set the faulty drive to "Offline" using the GUI.
I then used the Server for a day in DEGRADED mode, as expected. The next day I got a new drive, simply put it in and clicked "Replace". Silvering process started and completed after 3.5 hours.