ZFS replace disks by id

Question

By accident, I created a zpool using /dev/sda and so on. I knew this was a bad idea; I just wanted to test my setup. It turned out that it worked so well, I forgot about the device names and started to use my zpool as NAS, still with the sda, sdb, sdc and sdd. My operating system runs from sde. So my zpool is already filled with some data, around 16 TB to be precise. So I wonder whether it is possible to modify the existing pool to use the disks by id, and not by their sdX names?

My NAS has run continuously for about 1 year, and I have never rebooted. Could my zpool get destroyed if I reboot now and some disk names would change? (e.g., because I added some new disks afterwards)

I read that it may be possible using zpool export and zpool import -d /dev/disk/by-id. Will that cause some resilvering to start, or what exactly happens when I export and import? Since it is a lot of data, I would prefer if I don't have to copy my data around. It’s simply too much and would take days. My zpool runs in raidz2 configuration and the operating system is Debian.

This is what I get by zfs list all:

root@pve:~# zfs get all
NAME                                           PROPERTY                VALUE                    SOURCE
tank                                          type                    filesystem               -
tank                                          creation                Sat May 12 15:22 2018    -
tank                                          used                    1.00T                    -
tank                                          available               4.26T                    -
tank                                          referenced              981G                     -
tank                                          compressratio           1.02x                    -
tank                                          mounted                 no                       -
tank                                          quota                   none                     default
tank                                          reservation             none                     default
tank                                          recordsize              128K                     default
tank                                          mountpoint              /tank                    default
tank                                          sharenfs                off                      default
tank                                          checksum                on                       default
tank                                          compression             lz4                      local
tank                                          atime                   off                      local
tank                                          devices                 on                       default
tank                                          exec                    on                       default
tank                                          setuid                  on                       default
tank                                          readonly                off                      default
tank                                          zoned                   off                      default
tank                                          snapdir                 hidden                   default
tank                                          aclinherit              restricted               default
tank                                          createtxg               1                        -
tank                                          canmount                on                       default
tank                                          xattr                   on                       default
tank                                          copies                  1                        default
tank                                          version                 5                        -
tank                                          utf8only                off                      -
tank                                          normalization           none                     -
tank                                          casesensitivity         sensitive                -
tank                                          vscan                   off                      default
tank                                          nbmand                  off                      default
tank                                          sharesmb                off                      default
tank                                          refquota                none                     default
tank                                          refreservation          none                     default
tank                                          guid                    18018951160716445859     -
tank                                          primarycache            all                      default
tank                                          secondarycache          all                      default
tank                                          usedbysnapshots         100M                     -
tank                                          usedbydataset           981G                     -
tank                                          usedbychildren          47.5G                    -
tank                                          usedbyrefreservation    0B                       -
tank                                          logbias                 latency                  default
tank                                          dedup                   off                      default
tank                                          mlslabel                none                     default
tank                                          sync                    standard                 default
tank                                          dnodesize               legacy                   default
tank                                          refcompressratio        1.02x                    -
tank                                          written                 0                        -
tank                                          logicalused             1004G                    -
tank                                          logicalreferenced       997G                     -
tank                                          volmode                 default                  default
tank                                          filesystem_limit        none                     default
tank                                          snapshot_limit          none                     default
tank                                          filesystem_count        none                     default
tank                                          snapshot_count          none                     default
tank                                          snapdev                 hidden                   default
tank                                          acltype                 off                      default
tank                                          context                 none                     default
tank                                          fscontext               none                     default
tank                                          defcontext              none                     default
tank                                          rootcontext             none                     default
tank                                          relatime                off                      default
tank                                          redundant_metadata      all                      default
tank                                          overlay                 off                      default
tank@zfs-auto-snap_monthly-2019-05-13-1847    type                    snapshot                 -
tank@zfs-auto-snap_monthly-2019-05-13-1847    creation                Mon May 13 20:47 2019    -
tank@zfs-auto-snap_monthly-2019-05-13-1847    used                    0B                       -
tank@zfs-auto-snap_monthly-2019-05-13-1847    referenced              953G                     -
tank@zfs-auto-snap_monthly-2019-05-13-1847    compressratio           1.01x                    -
tank@zfs-auto-snap_monthly-2019-05-13-1847    devices                 on                       default
tank@zfs-auto-snap_monthly-2019-05-13-1847    exec                    on                       default
tank@zfs-auto-snap_monthly-2019-05-13-1847    setuid                  on                       default
tank@zfs-auto-snap_monthly-2019-05-13-1847    createtxg               6244379                  -
tank@zfs-auto-snap_monthly-2019-05-13-1847    xattr                   on                       default
tank@zfs-auto-snap_monthly-2019-05-13-1847    version                 5                        -

I tried to mount it with zfs mount -a, but that fails because the directory /tank is not empty – the other ZFS datasets are in there...

root@pve:~# zpool list -v
NAME   SIZE  ALLOC   FREE  EXPANDSZ   FRAG    CAP  DEDUP  HEALTH  ALTROOT
tank   5.44T   987G  4.47T         -     1%    17%  1.00x  ONLINE  -
  ata-WDC_WD20EFRX-68EUZN0_WD-WCC4M1KZNLPE  1.81T   326G  1.49T         -     1%    17%
  ata-WDC_WD20EFRX-68EUZN0_WD-WCC4M1YV1ADT  1.81T   329G  1.49T         -     1%    17%
  ata-WDC_WD20EFRX-68EUZN0_WD-WCC4M2CE10DJ  1.81T   332G  1.49T         -     1%    17%

score 9 · Accepted Answer · answered May 18 '19 at 13:06

9

The zpool won't get destroyed if the names of the disks change. The pool will very likely not import automatically, but data should not be destroyed. Unless there is a script or mechanism which doing things on blockdevices and pathes like /dev/sda are hardcoded and things run without sanity checks. But normally your data is safe.

Importing the pool with zpool import -d /dev/disk/by-id <pool-name> is also save to use. There is no resilvering needed and as far as I can tell the /etc/zfs/zpool.cache file is updated with the new paths and the ondisk metadata as well.

answered May 18 '19 at 13:06

Thomas

6,362

I just did zpool export tank and then zpool import tank -d /dev/disk/by-id. Wow. That gives me an error! a pool with that name already exists. I have a few subvolumes on my zpool, these show up in the zpool status, but the directory /tank, where normally my files are, is empty! zpool list however still shows that the space is actually used. But how do I get my files back? – T. Pluess May 18 '19 at 17:53
Do a zpool export -a and then show the output of zpool import -d /dev/disk-by-id without giving the name. Just to list the available pools. – Thomas May 18 '19 at 17:58
The message you are seeing a pool with that name already exists indicates that the pool was not exported correctly. Check with zpool list -v after you did a export to ensure that the pool was exported. – Thomas May 18 '19 at 18:04
zpool import -d /dev/disk-by-id shows: no pools available to import. However, zfs get all shows mounted: no – T. Pluess May 18 '19 at 18:06
The zfs command manages the filesets which are part of the pool. Use zpool list -v to list imported pools. The pool seems not to be exported. Use zpool export -a. – Thomas May 18 '19 at 18:08

ZFS replace disks by id

1 Answers1