Remove Log VDEV 22.12.3.3

[a.dresner]

Hello,

Added a mirror VDEV log, want to remove it. The log was supposed to be removable from what I read? Only option I see is to extend under storage/devices. Relative output

sudo zpool status -v
[sudo] password for admin:
pool: GALAXY
state: ONLINE
scan: scrub repaired 0B in 07:49:50 with 0 errors on Sun Oct 29 07:49:52 2023
config:

NAME STATE READ WRITE CKSUM
GALAXY ONLINE 0 0 0
raidz2-0 ONLINE 0 0 0
591ffe98-be66-40eb-b181-a81a09b99bf0 ONLINE 0 0 0
b29a275c-bc36-45bb-8f63-12c63f97549f ONLINE 0 0 0
de578d60-1ae4-4768-96ec-9a82235c3e71 ONLINE 0 0 0
2537b1eb-c975-471d-8a02-6b0578b142b9 ONLINE 0 0 0
4ddf5409-6838-4cf8-b209-f8b6200af834 ONLINE 0 0 0
03e4197a-524b-4383-ac23-875a7baae237 ONLINE 0 0 0
special
mirror-2 ONLINE 0 0 0
486b8455-b1ef-4fd5-b130-12b2fba823d2 ONLINE 0 0 0
548ff5f1-c105-4b6d-ad54-73fdc1e63cb4 ONLINE 0 0 0
logs
mirror-1 ONLINE 0 0 0
63ca2229-9871-4230-9fc8-071d79c84573 ONLINE 0 0 0
4e12728b-b849-485c-a6a1-fa14d68aa05f ONLINE 0 0 0

With an option in the GUI, would I need to remove via CLI? Would the correct commands be:

zpool status GALAXY
zpool offline GALAXY mirror-1
zpool remove GALAXY mirror-1

 

[Arwen]

I vaguely recall their being a bug in the GUI missing something related to non-data vDevs.

In theory, yes that command line should work. If not, you can remove one individual sub-mirror, then the other.

If you do remove the SLOG vDev from your pool, the GUI may not know it's gone. May need to either export & import your pool. Or reboot. Though it will probably be harmless to ignore it for a while.

 

[HoneyBadger]

I vaguely recall their being a bug in the GUI missing something related to non-data vDevs.

Correct, NAS-122781 to be specific, and the bug was squashed in 22.12.4

@a.dresner have you considered updating to the latest Bluefin release of 22.12.4.2? No need to switch trains to Cobia (23.10.x) yet if you aren't ready. If you'd rather not upgrade at this time, then the commands you laid out should handle it correctly from the shell.

Do note that regardless of version, your mirror-2 special vdev is there for good - you have a RAIDZ2 data vdev which will prevent other top-level data vdev removals including special.

 

[a.dresner]

I ran across NAS-122781 and upgraded but did not see the button... saw your msg and went back, same but refreshed my browser (safari) and the bottom appeared, log gone. =)

One follow up question if you don't mind, I had been trying to create a Metadata VDEV, I know it's there for good. But I have a Redundancy mismatch I hope I can correct?

As you know from above, I have:
1 x RAIDZ2 | 6 wide | 10.91 TiB
The special Metadata VDEV is currently:
1 x MIRROR | 2 wide | 110.28 GiB

Now that my log is cleared, I have 2 disks (110,28). Can I extend the special with 1 (or will it require 2 to match the 2 in the RAIDZ2?) disk to clear that redundancy mismatch?

Thank you

 

[asap2go]

I ran across NAS-122781 and upgraded but did not see the button... saw your msg and went back, same but refreshed my browser (safari) and the bottom appeared, log gone. =)

One follow up question if you don't mind, I had been trying to create a Metadata VDEV, I know it's there for good. But I have a Redundancy mismatch I hope I can correct?

As you know from above, I have:
1 x RAIDZ2 | 6 wide | 10.91 TiB
The special Metadata VDEV is currently:
1 x MIRROR | 2 wide | 110.28 GiB

Now that my log is cleared, I have 2 disks (110,28). Can I extend the special with 1 (or will it require 2 to match the 2 in the RAIDZ2?) disk to clear that redundancy mismatch?

Thank you

It's about the redundancy level.
A three way mirror can loose 2 drives and still be fine.
Same for raidZ2.
So if you add one drive to your mirror then the warning will go away.

 

[HoneyBadger]

Now that my log is cleared, I have 2 disks (110,28). Can I extend the special with 1 (or will it require 2 to match the 2 in the RAIDZ2?) disk to clear that redundancy mismatch?

Yes, but ensure that you "Extend" the existing vdev as mentioned by @asap2go - this will turn it from a 2-way to a 3-way mirror, able to tolerate an equal number of drive failures as your RAIDZ2 data vdev.

 

[asap2go]

Yes, but ensure that you "Extend" the existing vdev as mentioned by @asap2go - this will turn it from a 2-way to a 3-way mirror, able to tolerate an equal number of drive failures as your RAIDZ2 data vdev.

Is it possible to disable that warning in the GUI?
I have a similar configuration and I hate thinking of buying a new drive just to make the light go green...

 

[HoneyBadger]

Is it possible to disable that warning in the GUI?

Not at present; it's designed to warn against the idea of "any two disks can fail" in a RAIDZ2 with a meta-mirror, and someone thinking they're safe to remove two bays at a time.

We're always looking at ways to improve the clarity of information communicated by the UI; if you have a suggestion on how we can fix this, we'd love to hear it, either directly as a Jira ticket, a diagrammed mockup on the forums, some hacked-up CSS on a test system of yours, or even a direct PR to our Github.

 

[asap2go]

either directly as a Jira ticket, a diagrammed mockup on the forums, some hacked-up CSS on a test system of yours, or even a direct PR to our Github.

I would rather shoot my server than touch any Jira tickets in my free time :D
But thanks for the answer!
Guess I'll have my first three way soon.

 

[a.dresner]

Not at present; it's designed to warn against the idea of "any two disks can fail" in a RAIDZ2 with a meta-mirror, and someone thinking they're safe to remove two bays at a time.

We're always looking at ways to improve the clarity of information communicated by the UI; if you have a suggestion on how we can fix this, we'd love to hear it, either directly as a Jira ticket, a diagrammed mockup on the forums, some hacked-up CSS on a test system of yours, or even a direct PR to our Github.

First off...thank you, support has been great.

I get the point, it's pointing out a weak link so to speak. And given that specials like this are permanent and part of the overall structure of the pool..it makes sense. However I'm using Optane and the chance of blowing out both tires at the same time is pretty low and I can easily replace 1 stick if it goes bad rather quickly. Maybe have the option to turn off that warning might not be a bad idea? Agree to a pool death warning and then it's off.

It's very much a personal choice (in a home lab, in an office it shouldn't be a choice..add it) to add in the 3rd stick or ignore the warning. Obviously the responsible thing would be to add it in, it would also play nice against anyone's OCD.

The process starts here?

Not here?

 

[Etorix]

Definitely the first one.
Enter the second one and you're making a stripe of a mirror and a single drive as your "special" device… This could end badly.

 

[HoneyBadger]

Maybe have the option to turn off that warning might not be a bad idea? Agree to a pool death warning and then it's off.

It's very much a personal choice (in a home lab, in an office it shouldn't be a choice..add it) to add in the 3rd stick or ignore the warning. Obviously the responsible thing would be to add it in, it would also play nice against anyone's OCD.

I'll write it up; the are a couple other changes that have been requested to that screen, so no promises on how the engineering and UI/UX teams will prioritize though.

And as @Etorix says, definitely Option 1 there to "Extend" your special vdev.