he volume BASE-ZFS (ZFS) status is UNKNOWN

[vlky]

Hi Support,

I have created a pool "BASE-ZFS" in 4 harddrives A, B, C, D. Last week, the System prompt a waring as following. Each harddrive is 2TB and it is set RAID-5

I found that disk "D" was missing when I "View Disks". So I physical removed disk "D" from the server, and bought a new harddrive "E" and insert into the server then import the volume into the BASE-ZFS pool.

The warning is still prompting. What is the problem???
I am really really afraid that there is data lost!!!!!



I attached some screen dump for more information to you.

 

[cyberjock]

One of your disks is offline according to the zpool status you provided. Also, RAID5 is a hardware RAID type. You are running RAIDZ1. You might want to say RAIDZ1 next time to make the differentiation.

 

[vlky]

Yes, it should be raidz1.
But how do I update the status or use another disk replace that ???

 

[cyberjock]

http://wiki.freenas.org/index.php/Volumes#Replacing_a_Failed_Drive_or_SSD

 

[titan_rw]

Also, unless I'm seeing that wrong, you've got 4 disks in raid-z1 (one of which is missing), striped with another disk. If that last disk goes bad, you'll lose the entire pool. You may want to rethink your pool design, as you don't have redundancy over the entire pool.

 

[cyberjock]

Also, unless I'm seeing that wrong, you've got 4 disks in raid-z1 (one of which is missing), striped with another disk. If that last disk goes bad, you'll lose the entire pool. You may want to rethink your pool design, as you don't have redundancy over the entire pool.

He is correct. Your last disk isn't in the RAIDZ1 vdev, so if that disk fails your pool is done for.

 

[vlky]

Yea~ I replaced the disk for the RaidZ1.

Thanks!! BUT......

When I tried to extend the volume using "ZFS Volume Manager" (via Volume Layout).

I did something wrong and I forgot the procedure already......
The zpool's volume layout become very very strange as the screen dump..... (extra stripe volume)

How I can remove the stripe volume and no affected with the raidz1-0??
After I did that, I shutdown the system for avoiding any data was writer into the stripe volume /__\

Additional:
I studied many of many articles. I found that is there no-way/not possible to expand the RAIDZ-1 volume size(disk number) via adding hard-drives???

 

[cyberjock]

You can't remove the disk(which is now a vdev). Any vdev, once created cannot be removed or changed(with one exception for changing a vdev). That one exception is a single disk vdev can be converted into a mirrored vdev.

Your mistake is very common, which is one of many reasons why I created the noobie presentation over a year ago(link in my sig). Fixing the mistake is far from trivial, especially if you aren't making backups of your entire pool.

 

[vlky]

Thanks cyberjock very much!!
I saw your noobie section already XD
I understand what is the limitation to my use case, you help me a lot and I think I can keep this setting until I change my data server~

 

[cyberjock]

I'd highly recommend you mirror that single disk as soon as possible. If you lose that disk you lose the pool. And there are no data recovery tools to save you. :(

 

[vlky]

Mirror the stripe(ada3p2) ?
or mirror the whole raidz1-0?? ( means I have to buy 3 more hard-drives /__\ )

 

[cyberjock]

mirror ada3p2. You only need 1 more disk.

 

[vlky]

I believe in your professional recommendation.
But after I looked into some articles, I can not found the "Mirror" selection in ZFS Volume Manager. (Or there is other way to do???)

Any articles can I refer to ?? :)


One more
Am I falling into the situation like following??
So that I have to add one more disk(black) for ada3p2.
And once there is any write on BASE-ZFS, it will be randomly store to left or right side ?

 

[titan_rw]

Ideally, you would backup, and restore to a corrected zpool.

But if that's not possible immediately, then to avoid the single point of failure, you should add redundancy to the single disk vdev. I think this has to be done via the command line. A quick search shows this for converting single disk to mirror:

Code:

gpart create -s gpt ada0
gpart add -i 1 -t freebsd-swap -s 2G ada0
gpart add -i 2 -t freebsd-zfs ada0
zpool attach BASE-ZFS gptid/XXXXX ada0p2

Where XXXXX is the gptid from the single disk vdev.

Assuming that ada0 is what the new disk shows up as. And assuming you were using 2G (the default) for swap.

And yes, new data will be spread across your old raidz1 vdev, and the new (currently single disk) vdev. However, because of how zfs works, loss of the (currently single disk) vdev will result in loss of all data, not just recently added data.