Quotas: Name not found

[Zekas]

Hi everyone,

I just discovered that on some of my datasets there are still quotas for UID/GID's that don't exist anymore.
How can I figure out which data this is as well as fix the quota group/user list?
I've tried a recursive chown on one of the datasets to make sure all files are assigned to the correct (existing) user, but the quota still show objects and DQ used.
FYI: I've never actually used the quota functionality itself either.

Thanks in advance for any help!

 

[anodos]

Hi everyone,

I just discovered that on some of my datasets there are still quotas for UID/GID's that don't exist anymore.
How can I figure out which data this is as well as fix the quota group/user list?
I've tried a recursive chown on one of the datasets to make sure all files are assigned to the correct (existing) user, but the quota still show objects and DQ used.
FYI: I've never actually used the quota functionality itself either.

Thanks in advance for any help!

View attachment 48409

Can you file a bug report on our Jira instance? jira.ixsystems.com

 

[Zekas]

I think I could, have to see about making an account for it later since it needs one. Just to be sure: is it a bug though? The UID/GID's have been actual groups and users in the past, but they have been deleted.

 

[anodos]

I think I could, have to see about making an account for it later since it needs one. Just to be sure: is it a bug though? The UID/GID's have been actual groups and users in the past, but they have been deleted.

I believe the numeric ID should be displayed even if it does not resolve to a name.

 

[Zekas]

I believe the numeric ID should be displayed even if it does not resolve to a name.

If I click the 'down' arrow, I can see and edit the G/U)ID (which is 1002 in my example screenshot). So that's not the problem. It wasn't really one of my questions either :)

 

[Zekas]

To clarify my previous post, here's another screenshot with the dropdown activated:

 

[anodos]

Sorry, misread your post. There is a general issue with how these are being filtered (probably should not include ones that do not have quotas set). I filed ticket in jira for that. In general the API here is a thin wrapper around "zfs userspace".
```
zfs-userspace – Displays space consumed by, and quotas on, each user or
group in the specified filesystem or snapshot.
```
So you're seeing space used by files owned by id 1002.

 

[Zekas]

Alright, I understand that, but that still doesn't answer my questions. How can I find out what files they are? Like I said in my original post, I did a full recursive chown on the entire dataset, how can there still be files/objects belonging to UID 1002? How do I fix it?

 

[Zekas]

I also tried a find for the UID/GID listed on the quota page, but no results:

root@nas:~ # find /mnt/personal/[redacted]/ -user 1002
root@nas:~ #
root@nas:~ # find /mnt/personal/[redacted]/ -group 1001
root@nas:~ #

 

[anodos]

I also tried a find for the UID/GID listed on the quota page, but no results:

root@nas:~ # find /mnt/personal/[redacted]/ -user 1002
root@nas:~ #
root@nas:~ # find /mnt/personal/[redacted]/ -group 1001
root@nas:~ #

Does the entry appear if you run zfs userspace <pool>/<dataset> from CLI?

 

[Zekas]

It does:

root@nas:~ # zfs userspace personal/[redacted]
TYPE NAME USED QUOTA OBJUSED OBJQUOTA
POSIX User [redacted] 2.61G none 161K none
POSIX User [redacted] 9.18M none 18.3K none
POSIX User [redacted] 758G none 527K none
POSIX User 1002 80.2M none 120K none

 

[Zekas]

Hi Anodos, any other ideas after the provided output?