ZFS Pool Degraded - Cleared - Waiting for error

[jgiddens]

Hi All,

Over the last 6 months I have had my ZFS pool show up as degraded twice. It consists of 3 mirrors of 2 4Tb drives each, running on a R720 dell server. The first time I did a zfs clear command to see if the problem reoccured. It did, but only after a couple of months. The second time I cleared again and decided I would replace the two drives that seemed to have the issue. It is one mirrored pair that is marked as degraded. I set about finding replacement drives but took my time since it didnt seem too pressing. Its now been a couple of months and it has not reoccured but I have 2 new drives ready to replace. The problem is I cannot remember which pair had the issue. I can just wait around for it to reoccur, but I would prefer to change the drives out now.

Suggestions on how I can figure out which drives previously had the issues?

Also, once I do figure out which drives, any advice on how to tie those drive names to serial numbers so I know which ones to pull from my chassis?

Thanks!

 

[leesteken]

If a mirror is degraded, only one of the disks had an error and it could be recovered from the other. Otherwise, the status would have been failed instead of degraded.
(Unless there were multiple errors and luckily each separate error had correct data on the other disk.) Just saying that you maybe don't have to replace both drives.
I would expect that a ls -l /dev/disk/by-id/ (maybe with the help of ls -l /dev/disk/by-path/) can help you locate drives by serial number.

 

[Dunuin]

I would expect that a ls -l /dev/disk/by-id/ (maybe with the help of ls -l /dev/disk/by-path/) can help you locate drives by serial number.

Its also best to add/replace the drives by "/dev/disk/by-id/yourdisk" instead of for example "/dev/sda". Then a zpool status will list the disks with their unique ID which usually contains the disks serial number so its easier to see which disk to replace if it is causing problems.

 

[spacedragonspirit]

I use https://github.com/AnalogJ/scrutiny
for monitoring my drives from time to time. Maybe it helps you to determine your perilously failed disk.

 

[jgiddens]

If a mirror is degraded, only one of the disks had an error and it could be recovered from the other. Otherwise, the status would have been failed instead of degraded.
(Unless there were multiple errors and luckily each separate error had correct data on the other disk.) Just saying that you maybe don't have to replace both drives.
I would expect that a ls -l /dev/disk/by-id/ (maybe with the help of ls -l /dev/disk/by-path/) can help you locate drives by serial

Thanks... I am also enlarging the pair from 4Tb to 8Tb, so I am seizing on the opportunity created by this degraded situation.

Any idea what logs would show the previous errors?