Running Raid-Z2 Degraded?

[doodlebob]

I am running 6 x 4TB WD REDs in Raid-Z2. One of my disks has developed "read failure" when running my scheduled Extended SMART test and I have initiated an RMA.

I've got the disk removed and sent off and I don't have a cold spare.

My question is: Would I be mistaken and running great risk to continue operating in a DEGRADED state for the next week while my drive is away? I'll be doing a badblocks burn-in on the replacement as well so I would be running for possibly two weeks degraded.

I wanted to get others opinions on what they'd do as this is my first foray into the RMA process and utilizing the fault-tolerance of Raid-Z2.

Thanks!

 

[Spearfoot]

I am running 6 x 4TB WD REDs in Raid-Z2. One of my disks has developed "read failure" when running my scheduled Extended SMART test and I have initiated an RMA.

I've got the disk removed and sent off and I don't have a cold spare.

My question is: Would I be mistaken and running great risk to continue operating in a DEGRADED state for the next week while my drive is away? I'll be doing a badblocks burn-in on the replacement as well so I would be running for possibly two weeks degraded.

Please save me from myself as this is my first foray into the RMA process and utilizing the fault-tolerance of Raid-Z2.

Thanks!

As with so much of life, the answer is "It Depends".

You would have to lose two more disks before you would lose the data in your RAIDZ2 pool... This is statistically unlikely, so it's relatively safe to run in a degraded state for a brief time. But there is indeed a real risk as losing two disks is not entirely unheard-of.

The safest course, if you can live without access to your data until you have a tested replacement disk, would be to leave the system powered down until you can replace the failed disk.

 

[danb35]

Pretty much what @Spearfoot said. I've paid the few bucks for advance RMA service before--it gets me the new disk faster, and also gives me a shipping container and paid shipping label for the dead disk, so the actual cost to me is pretty low. May be something to consider in the future. But aside from that, your risk is pretty low, though it's certainly higher than it was before the disk died.

 

[Stux]

I generally run out and buy a new disk, then the RMA becomes a spare.

RMA returned drives are unreliable refurbished rubbish anyway.

It depends on how critical it is to you.

Refresh your backup, and replace the failed drive as soon as is practical.

If your backup is up2date, then so what?

...

BUT most of the time that this happens to me the drives are out of warranty, which makes the decision easier... and my feelings about refurbishedRMA replacement drives may be coloured by very bad experiences with Seagate 1.5 and 3TB drives ;)

(yeay for >100% failure rates)

 

[danb35]

very bad experiences with Seagate 1.5 and 3TB drives ;)

...the optimal storage configuration of which is buried as deep as possible in the nearest landfill.

Edit: Credit @DrKK with this clarity; I'm not nearly that witty.

 

[Stux]

That's where they all are now

 

[nojohnny101]

I have run my box in raidz2 in a degraded state. all good advice above. I was a cheap skate and didn't pay for advanced RMA. I did seem to suffer lower performance on the pool while this was the case.

I took my chances though because I couldn't go with the server being down for a week while I waited AND I have an offsite backup (another FreeNAS) box was ready to go (and up to date) if things really went bad. You have to weigh your options.

 

[rs225]

In my opinion, the primary reason for raidz2/3 is exactly so you can run the pool like this. The reasons everybody else uses it are all the 'wrong reasons': Fear of double disk failure, and not making backups.

 

[Jailer]

In my opinion, the primary reason for raidz2/3 is exactly so you can run the pool like this. The reasons everybody else uses it are all the 'wrong reasons': Fear of double disk failure, and not making backups.

No the reason most of us run it is we care about our data.

If you plan on running a degraded pool why set it up with parity in the first place?

 

[Z300M]

I generally run out and buy a new disk, then the RMA becomes a spare.

RMA returned drives are unreliable refurbished rubbish anyway.

No, Refurbished drives have undergone more extensive testing than the ones that come brand new in a fancy carton.

The replacement I received for one of my 5-year-warranty Seagates was a model that would normally have had only a 2-year warranty but in fact came with the balance of the 5-year warranty and outlasted that. And even then its only SMART error was very minor but I replaced it anyway.

 

[rs225]

No the reason most of us run it is we care about our data.

If you plan on running a degraded pool why set it up with parity in the first place?

Availability. It lets you continue operation during an event that would otherwise cause downtime and a restoration from backup. Higher levels for longer periods of degraded operation. But RAID is not a backup. The only way to care for your data is to make independent copies of it.

http://www.taobackup.com/

 

[nojohnny101]

I agree with @rs225 No one wants to run a degraded pool, but I personally can't always take care of the dead disk right away (recently one died while I was on the road for 4 weeks). I was still able to access my media and my server never went down. I didn't lose sleep over it because the risks of losing 2 more disks before I could get it it were small enough and I had an offsite backup.

Redundancy in my mind, for my circumstances, simply buys me time and allows me to take care of other things in my life that are more important instead of having to be at the beckon call of my server.

To each his/her own!