Sometimes we have to add disks from another system, or disks that are non-zeroed spares, to replaced a failed disk in a system either without a spare or with more than one failed in a single aggregate.
This is fairly simple, first, your replacement disk needs to be of the same speed as the ones contained in the aggregate. DataONTAP doesn't allow mixed speeds for performance reasons by default. You can change the setting to allow mix-speed drives in the same aggregate by changing raid.rpm.fcal.enable to allow mixed but this isn't recommended.
First replace the failed disk with the replacement disk.
If the replacement disk already is a member of a foreign volume or aggregate it will show up appended with a "(1)", for example, if your replacement disk is from a system's vol0 it will show up in the new system as vol0(1) and will be brought in "offline" as noted as "foreign." This is helpful for the next step.
Locate the new disk in the system, if it is a member of a foreign volume or aggregate list it with: vol status or aggr status and note the volume or aggregate name it is a member of, e.g. vol0(1).
rsyncnas-ancf> vol status
Volume State Status Options
vol0(1) offline raid_dp, flex fs_size_fixed=on
foreign degraded
vol0 online raid_dp, flex root
irt online raid_dp, flex nosnapdir=on, fs_size_fixed=on
ndfd online raid_dp, flex nosnapdir=on, fs_size_fixed=on
First you need to destroy the volume the new disk was previously a member of,
> vol destroy vol0(1)
BE VERY CAREFUL HERE ... double check to make sure that you enter the foreign volume and not the system's root volume.
Once the volume is destroyed, the disk will now become a spare, you can list the spares on the system via: vol status -s or aggr status -s. At this point you will need to zero the spare disk with:
> disk zero spares
Once the disk zeroes it is ready to be readded to the system. Assuming previously you did not have a spare. If you lost a dparty disk, the RAID will have degraded down from a raid_dp to raid4 via DataONTAP automechanisms. You can change the RAID on that aggregate from which the failed disk came via:
> aggr options aggr1 raidtype raid_dp
DataONTAP will automatically begin rebuilding/reconstructing the RAID with the spare you just added.
If a disk failed when you were in the process of failing over the resources from one controller to another, and you failed back before the disk was reaplced, sometimes DataONTAP will assign that new spare to the controller that originally owned the failed disk, but the reconstruct will have occurred on the partner node when the cluster was in takeover mode. Therefore, the partner node will have one fewer spares, but when you insert the new disk it will be assigned to the node to which the original failed disk belong, and so that partner will have one too many spares.
To reassign a spare disk to a partner controller:
> disk assign 1a.10.11 -o nas2-nhcn -f
Assuming 1a.10.11 is a spare and the partner nodename is nas2-nhcn. You can use the partner systemID with the -s option.
No comments:
Post a Comment