Unable to create a new share on the rockstor_rockstor pool and quota error

[razaborg]

Hi,

This is my setup :
2 pools:

• 1 rockstor_rockstor pool with 1 SSD, only the OS
• 1 data_volume pool with 2 HDD, RAID 1, everything else

I wanted to create a new share on my first pool (rockstor_rockstor), to see how faster this share could be on an SSD drive...

But this error showed-up :

Traceback (most recent call last):
  File "/opt/rockstor/src/rockstor/rest_framework_custom/generic_view.py", line 41, in _handle_exception
    yield
  File "/opt/rockstor/src/rockstor/storageadmin/views/share.py", line 187, in post
    share_pqgroup_assign(pqid, s)
  File "/opt/rockstor/src/rockstor/fs/btrfs.py", line 889, in share_pqgroup_assign
    mnt_pt = mount_root(pool)
  File "/opt/rockstor/src/rockstor/fs/btrfs.py", line 930, in qgroup_assign
    # and deal by returning False so our caller moves on.
CommandException: Error running a command. cmd = /sbin/btrfs qgroup assign 0/1177 2015/15 /mnt2//rockstor_rockstor. rc = 1. stdout = ['']. stderr = ['ERROR: unable to assign quota group: Invalid argument', '']

Fine. I then tried to create the same share on my other pool (2 HDD, RAID 1), just to see what happen... And the share was just created successfully...

I tried to remove the Quota on the rockstor_rockstor, and then, a new error showed-up :

Traceback (most recent call last):
  File "/opt/rockstor/src/rockstor/rest_framework_custom/generic_view.py", line 41, in _handle_exception
    yield
  File "/opt/rockstor/src/rockstor/storageadmin/views/pool.py", line 351, in put
    'level: %s' % raid_level)
TypeError: %d format: a number is required, not unicode

This is clearly a Python programming issue. But that does not explain the first error. :(

[phillxnet]

@razaborg Thanks for reporting your findings.
Using latest stable I am able to create a share on the system pool (usually rockstor_rockstor) with quotas enabled or disabled. Could you report the displayed Rockstor version (top right) and confirm it with:

yum info rockstor

as root user.

Thanks.

What look odd is that a code comment is showing in your Traceback:

# and deal by returning False so our caller moves on.

Have you hand edited the code or edited the traceback before posting as I haven't see this before?

As to your second point "I tried to remove the Quota on the rockstor_rockstor," a quota disabled state is only supported in the very latest stable release channel currently. Prior to that there are multiple breakages with quotas disabled.

Also to your: "This is clearly a Python programming issue. But that does not explain the first error. :("
agreed and there is an ongoing effort to resolve this type of error going forward via:
"root pool edit message formatting type error" #1847
which includes reviving our unit test to help prevent this going forward.
It's actually what I'm currently working on.

[razaborg]

Here is the output of yum info rockstor:

Paquets installés
Nom                 : rockstor
Architecture        : x86_64
Version             : 3.9.2
Révision            : 19
Taille              : 77 M
Dépôt               : installed
Depuis le dépôt     : Rockstor-Stable
Résumé              : RockStor -- Store Smartly
Licence             : GPL
Description         : RockStor -- Store Smartly

For the record, I subscribed for the stable update releases plan, so I should have the latest stable version available : RockStor 3.9.2-19

And no, I didn't either edited manually the code nor the traceback before posting it here. I did it again to check and the error still shows up like that.
It's strange and I haven't ever seen that before either. Here is a copy of this part of the code:

def qgroup_destroy(qid, mnt_pt):
    cmd = [BTRFS, 'qgroup', 'show', mnt_pt]
    try:
        o, e, rc = run_command(cmd, log=True)
    except CommandException as e:
        # we may have quotas disabled so catch and deal.
        emsg = "ERROR: can't list qgroups: quotas not enabled"
        if e.err[0] == emsg:
            # we have quotas disabled so can't destroy any anyway so skip
            # and deal by returning False so our caller moves on.
            return False
        # otherwise we raise an exception as normal
        raise e
    for l in o:
        if (re.match(qid, l) is not None and l.split()[0] == qid):
            return run_command([BTRFS, 'qgroup', 'destroy', qid, mnt_pt],
                               log=True)
    return False

[phillxnet]

Closing for now as this relates to a now old and legacy base OS version of Rockstor.

@razaborg If you can reproduce this issue on our new "Built on openSUSE" base derived form one of our installers do please re-open this issue. We have had quite a few in-code and upstream fixes in the interim, plus of course the change form CentOS to openSUSE so hopefully things should be far better for your in this respect now.