Diskpool Degraded state - disk removed

[georgelza]

so curveball my side.

I had 2 drives, /dev/sdd and /dev/sde that now and then throw errors (they each belong to different disk pools, running 3 way RAIDz1), they would normally clear themself and system just happily continues, nothing new on these 2 drives... was starting to think I need to buy some new drives and replace them...

Tonight, suddenly got an alert, drive thats never thrown errors failed, was removed by TrueNAS from the disk pool.

All my drives are Seagate Ironwolf 4TB HDD's...

so, ye, it seems i am immediately in for a new replacement drive... looking future planning, need to extend the space. do I replace with a 4TB drive... or a 9TB, knowing I will only use for now 4TB, with the idea to do some musical chairs on tis disk pool once all 3 HDD's are 8TB.

and then again, look left, expecting a problem at some point and something smacks you from the right where you did not expect it coming from.

Murphy at work.

G

 

[georgelza]

if i was to stick to a 4TB HDD

The Seagate Ironwolf comes in ST4000VN006 and ST4000VN008 models.

My disk pool is build out of ST4000VN008 models.
would there be a problem if I replaced the failed drive with a ST4000VN006

G

 

[georgelza]

Found this:
The main differences are the
VN006 spins at 5400rpm
VN008 spins at 5900rpm

and the

VN006 has 256mb cache as opposed to
VN008 has 64mb

G

 

[sretalla]

I doubt it will cause any notable change.

The reality of it is that the pool will slow to the speed of its slowest disk, so if the new disk is somehow not able to perform as well as the existing ones, your pool performance may degrade (probably hard to notice if it does though).

 

[georgelza]

Think these 2 models are so close.

Can we discuss the other question. My diskpool is based on 4Tb HDDs atm… I need to look at increasing the size. Can I start future planning by replacing this disk with 8Tb unit. Knowing I might only use 4Tb on it for a while.
G

 

[sretalla]

Can we discuss the other question. My diskpool is based on 4Tb HDDs atm… I need to look at increasing the size. Can I start future planning by replacing this disk with 8Tb unit. Knowing I might only use 4Tb on it for a while.

Sure, I have done that in the past with no issue. You just don't get to the additional capacity until all disks are upgraded to the new size (but it seems you understood that already, so fine).

 

[georgelza]

That’s my understanding also.
Thanks for confirmation.
This way can replace drive/restore resilience… but at the same time put 1 of 3 larger disks down… to be followed by a second and then a third at which point I can then grow the pool.
G

 

[georgelza]

Can find this drive : ST10000NM002G for pretty much the same price as: Seagate Ironwolf Pro ST8000NE001

Please Comment. See my signature for the LSI controller it will plug into.

G

 

[sretalla]

The Exos drives are used by others in the forum as far as I've seen, no specific notes on them.

I have no personal experience with those, so can't give much guidance.

 

[georgelza]

Hope someone can comment, but that 10 Tb exo is same cost as smaller 8Tb…

G

 

[georgelza]

sretalla, little update​

I decided to buy 3 x 8TB Seagate Ironwood drives. Diskpool:"tank", will play musical chairs and replace one drive, rebalance, then pull another/replace with larger, thus swopping out the bad drive and the 2 good drives/allowing it to sync between, giving me a double current size disk pool. with my old datasets on this pool, i have another diskpool ("bunker") based out of another 3 x 4TB HDD. I will move the dataset located on that also onto this ("tank") pool, then take the 2 good / old 4TB drives from this "tank" pool and that one and create a new 5 way RaidZ2 pool ("bunker"), followed by moving some datasets back onto that older HDD based pool, but based on Z2.

G

 

[georgelza]

once the drives drives arrive.... if I can ask you to help me through the process, would appreciate it.

G

 

[georgelza]

sretalla,​

does TrueNAS expect naming persistenace on device names.

aka, if i moved HDD's around on the cable endpoints patched into y controller, then their /dev/sd* will change, know some storage platforms don't mind as they ready physical params and then match that together based on sun headers and older storage platforms could not handle this change ?

G

 

[sretalla]

No. gptid is used and is on the drive itself so can't be confused.

 

[georgelza]

No. gptid is used and is on the drive itself so can't be confused.

Awesome. Thanks for confirmation.
As the diskpool is degraded and the faulty drive is not even showing up anymore, assume I can just remove it, there is no pre work to be done…
Once new drive is connected accept there will be commands to be issued to tell it to use the new drive, as a first step.
G

 

[sretalla]

As the diskpool is degraded and the faulty drive is not even showing up anymore, assume I can just remove it, there is no pre work to be done…

If it's that far gone, nothing much you can do.

Once new drive is connected accept there will be commands to be issued to tell it to use the new drive, as a first step.

The disk replacement process in the documentation has the steps laid out.

/scale/scaletutorials/storage/disks/replacingdisks/

Those processes often mention to offline the disk first. Note that Faulted is also an offline state.

 

[georgelza]

My faulty disk is not even showing on the list of disks. it's completely gone...

thanks for the link above, will read it and get ready.

G

 

[georgelza]

please see other thread where I tagged you.

First drive arrived, thinking I'm going to push it onto the disk thats showing disk errors in bunker, that is now throwing errors every night, and then replace the 2nd drive that is throwing alerts, luckily not showing errors yet, followed by the 3rd, which will result in this disk pool going from 3 x 4TB to 3 x 8TB, which is then big enough for this/Bunker requirement, but also able to handle the storage from Tank (where I have that degraded stat),
I then want to relocate the datasets from tank onto bunker, releasing the 2 good 4TB drives, at which point it will be figuring out whats good and whats not, and then rebuild a new Tank, after which I will move tanks datasets back.

might have to buy another 1 or 2 x 4TB HDD's at the time, to make it 4 or 5 wide RaidZ2.

Comment with action plan ?

G

 

[sretalla]

As mentioned in the other thread, can I see the output from zpool status Tank Bunker

I'm trying to make sense out of what you're saying you want to do and maybe I'm missing something.

 

[georgelza]

See response that side.

G