How to replace a failed disk in a ZFS mirror
26th November, 2018
I recently built a new file server for my media needs at home. Something I’ve been thinking about doing for literally years. I chose to go with ZFS as the storage technology after having used Linux software RAID for many years. I went with a mirrored setup for a lot of the reasons outlined in this article - performance, simplicity, and in particular, easy recovery from disk failures.
This is the setup I ended up with according to zpool status.
$ zpool status
pool: storage
state: ONLINE
scan: none requested
config:
NAME STATE READ WRITE CKSUM
storage ONLINE 0 0 0
mirror-0 ONLINE 0 0 0
ata-WDC_WD80EFZX-68UW8N0_VJHDBDGX ONLINE 0 0 0
ata-WDC_WD80EFAX-68KNBN0_VAGASE7L ONLINE 0 0 0
mirror-1 ONLINE 0 0 0
ata-WDC_WD80EFZX-68UW8N0_VJHD6BAX ONLINE 0 0 0
ata-WDC_WD80EFAX-68KNBN0_VAGA5BPL ONLINE 0 0 0
mirror-2 ONLINE 0 0 0
ata-WDC_WD80EFZX-68UW8N0_VJHD982X ONLINE 0 0 0
ata-WDC_WD80EFAX-68KNBN0_VAG9X8YL ONLINE 0 0 0
errors: No known data errors
Well, no sooner had I completed the ZFS setup (a very straightforward process) than one of my disks started reporting SMART errors. I don’t think a disk that is weeks old should do this, so I decided to start the RMA process.
And this is how I replaced the disk.
Replacing the disk
I started by physically removing the old disk, and replacing with a brand new
one. I originally setup my pool using the disk id from /dev/disk/by-id/, so
identifying the failed disk was very easy as the serial number is part of the
device name
Once I started back up, I ran zpool status and saw this output.
$ zpools status
pool: storage
state: DEGRADED
status: One or more devices could not be used because the label is missing or
invalid. Sufficient replicas exist for the pool to continue
functioning in a degraded state.
action: Replace the device using 'zpool replace'.
see: http://zfsonlinux.org/msg/ZFS-8000-4J
scan: none requested
config:
NAME STATE READ WRITE CKSUM
storage DEGRADED 0 0 0
mirror-0 ONLINE 0 0 0
ata-WDC_WD80EFZX-68UW8N0_VJHDBDGX ONLINE 0 0 0
ata-WDC_WD80EFAX-68KNBN0_VAGASE7L ONLINE 0 0 0
mirror-1 ONLINE 0 0 0
ata-WDC_WD80EFZX-68UW8N0_VJHD6BAX ONLINE 0 0 0
ata-WDC_WD80EFAX-68KNBN0_VAGA5BPL ONLINE 0 0 0
mirror-2 DEGRADED 0 0 0
ata-WDC_WD80EFZX-68UW8N0_VJHD982X ONLINE 0 0 0
18311740819329882151 UNAVAIL 0 0 0 was /dev/disk/by-id/ata-WDC_WD80EFAX-68KNBN0_VAG9X8YL-part1
errors: No known data errors
ZFS noticed that it had a missing disk, and was now in a DEGRADED state, but
crucially, everything was still working and available.
The next step was to find out what the new device is called. I did this by
running ls -1 /dev/disk/by-id/ and seeing which disk was new.
$ ls -1 /dev/disk/by-id/ | grep ata
ata-WDC_WD80EFAX-68KNBN0_VAGA5BPL
ata-WDC_WD80EFAX-68KNBN0_VAGA5BPL-part1
ata-WDC_WD80EFAX-68KNBN0_VAGA5BPL-part9
ata-WDC_WD80EFAX-68KNBN0_VAGASE7L
ata-WDC_WD80EFAX-68KNBN0_VAGASE7L-part1
ata-WDC_WD80EFAX-68KNBN0_VAGASE7L-part9
ata-WDC_WD80EFAX-68LHPN0_7HJSWL7F
ata-WDC_WD80EFZX-68UW8N0_VJHD6BAX
ata-WDC_WD80EFZX-68UW8N0_VJHD6BAX-part1
ata-WDC_WD80EFZX-68UW8N0_VJHD6BAX-part9
ata-WDC_WD80EFZX-68UW8N0_VJHD982X
ata-WDC_WD80EFZX-68UW8N0_VJHD982X-part1
ata-WDC_WD80EFZX-68UW8N0_VJHD982X-part9
ata-WDC_WD80EFZX-68UW8N0_VJHDBDGX
ata-WDC_WD80EFZX-68UW8N0_VJHDBDGX-part1
ata-WDC_WD80EFZX-68UW8N0_VJHDBDGX-part9
The new disk is the one on line 8 - ata-WDC_WD80EFAX-68LHPN0_7HJSWL7F. It
stands out in this example as all the other disk serial numbers start with “V”.
I now needed to tell ZFS to replace the missing disk with this one.
sudo zpool replace -f storage 18311740819329882151 /dev/disk/by-id/ata-WDC_WD80EFAX-68LHPN0_7HJSWL7F
ZFS automatically started the resilvering process (copying data to the new disk). I wasn’t sure how long that would take…
$ zpool status
pool: storage
state: DEGRADED
status: One or more devices is currently being resilvered. The pool will
continue to function, possibly in a degraded state.
action: Wait for the resilver to complete.
scan: resilver in progress since Thu Nov 15 17:01:06 2018
7.97G scanned out of 7.51T at 233M/s, 9h22m to go
2.56G resilvered, 0.10% done
config:
NAME STATE READ WRITE CKSUM
storage DEGRADED 0 0 0
mirror-0 ONLINE 0 0 0
ata-WDC_WD80EFZX-68UW8N0_VJHDBDGX ONLINE 0 0 0
ata-WDC_WD80EFAX-68KNBN0_VAGASE7L ONLINE 0 0 0
mirror-1 ONLINE 0 0 0
ata-WDC_WD80EFZX-68UW8N0_VJHD6BAX ONLINE 0 0 0
ata-WDC_WD80EFAX-68KNBN0_VAGA5BPL ONLINE 0 0 0
mirror-2 DEGRADED 0 0 0
ata-WDC_WD80EFZX-68UW8N0_VJHD982X ONLINE 0 0 0
replacing-1 DEGRADED 0 0 0
18311740819329882151 UNAVAIL 0 0 0 was /dev/disk/by-id/ata-WDC_WD80EFAX-68KNBN0_VAG9X8YL-part1
ata-WDC_WD80EFAX-68LHPN0_7HJSWL7F ONLINE 0 0 0 (resilvering)
errors: No known data errors
The resilvering completed in 5 hours and 53 minutes. A figure I’m very satisfied with. In this mirrored setup the data is at risk whilst resilvering completes, so the quicker, the better.
$ zpool status
pool: storage
state: ONLINE
scan: resilvered 2.50T in 5h53m with 0 errors on Thu Nov 15 22:54:41 2018
config:
NAME STATE READ WRITE CKSUM
storage ONLINE 0 0 0
mirror-0 ONLINE 0 0 0
ata-WDC_WD80EFZX-68UW8N0_VJHDBDGX ONLINE 0 0 0
ata-WDC_WD80EFAX-68KNBN0_VAGASE7L ONLINE 0 0 0
mirror-1 ONLINE 0 0 0
ata-WDC_WD80EFZX-68UW8N0_VJHD6BAX ONLINE 0 0 0
ata-WDC_WD80EFAX-68KNBN0_VAGA5BPL ONLINE 0 0 0
mirror-2 ONLINE 0 0 0
ata-WDC_WD80EFZX-68UW8N0_VJHD982X ONLINE 0 0 0
ata-WDC_WD80EFAX-68LHPN0_7HJSWL7F ONLINE 0 0 0
errors: No known data errors
ZFS is easy to setup and use for the most part. It feels solid. Stable. If all disk replacements are this easy I will be very happy.