CIFS quota reporting file sizes/file count incorrectly

Question

I'm having a frustrating problem with Celerra 5.5. I have an open support ticket, but they are equally as stumped. I'm hoping someone here has run into this problem before and has a solution.

I am sharing out a filesystem via CIFS to contain our Windows users home directory (i.e. [cifs_server]/users/(username)). This works fine, but I want to enable user quotas on this file system.

The problem is that Windows and the Celerra disagree as to the number of files and thier sizes. For example, in my personal share, Celerra says I am using 110 GB of storage, while Windows reports that I'm using 1.79GB. A random sample of 10 other users indicates the same problem for other users as well.

I've made sure that my server_param is measuring "filesize", and I've unmount/remounted the file system to force recalculation. So far, nothing has work to bring the two numbers into accordance with each other.

There are no tree quotas, so user quotas are only affecting the single UID/SID (as far as I know).

Also, if I cannot always reliably see the "quota" tab in Windows from these mapped drives - sometimes I have to remove a mapping, then re-map it to get the quota tab to appear. And, when doing this, bringing up the "properties" tab for the mapped drive takes a very, very long time. I had been hoping that our helpdesk folks could manage quotas via Windows in this way, without needing to log into the Celerra or use the Web GUI.

Why are the file counts and file sizes different in Windows vs Celerra and what can I do about it?

dynamox · Answer

the problem i experienced is very similar to yours ..only in my case i was using tree quotas. While tree quota reported one value ..windows explorer reported another. The lab ran some kind of script to sync them up, i am surprised they are not aware of this script.

stegad · Answer

Thanks! Do you recall the name of the script or who you spoke to? So far, EMC doesn't know what I'm referring to by 'the lab'.

stegad · Answer

Was the script a one time thing, or did you need to run in regularly to fix the problem?

dynamox · Answer

one time thing.

BillStein-Dell · Answer

I'm in the lab, and I don't know of this script.Do you have the SR number from this incident that I can reference? Feel free to e-mail it directly to me if you don't want to post it.v/r, bill

dynamox · Answer

hang tight ..i will email my CE

stegad · Answer

Hi Bill,Thanks for looking into this. I appreciate it. I also have this open on SR#22425874. I'm really hopefull that this script will solve my problem. I based my entire migration to the SAN around the idea of implementing user quotas, so to discover that they aren't working correctly was a big disappointment. I've had this case open for awhile and so far nobody has been able to figure out why this is happening. I'm really counting on this mystery script to save the day.

richcrompton · Answer

stegad -Are the values being reported from Windows the actual values (used/available) for the entire file system on which your user home directories reside?

stegad · Answer

Sort of, if I follow you. Windows Explorer shows the size of the entire file system for each mapped drive, yes. So, for exampe, the mapped drive to my share reports a total size that equals the entire storage pool for that filesystem, instead of my quota. That's only a minor inconvenience.

The problem is that if I select "properties" of that mapped drive (and IF the quota tab decides to show up), the sizes reported for each share do not match what is actually there. Or if I lookup quota info from either the celerra CLI or web GUI.

The strange thing is that discrepancy is not consistent. My share shows the largest variance (110GB used in celerra vs. 1.7gb of data actually there), but other shares are off by a different, but significant, amount (i.e. Celerra reports 850 mb used when actually there's only 240 mb of "stuff").

richcrompton · Answer

Is the cifs.useQuotaInFsStat server parameter set? What's the output of 'server_param server_2 -facililty quota -info useQuotaInFsStat'?

BillStein-Dell · Answer

Did you recently change quota policies from blocks to filesize?

If you have not done this, I would highly suggest doing so. A policy of blocks calculates disk usage in terms of usage of 8k blocks and counts all files (including directories and symbolic links). A policy of filesize works more like Windows, calculating quotas only based on the filesize. CIFS quotas should only use the filesize param.

If you have changed the policy, did you reboot the Data Mover? Following a reboot, the Data Mover will recalculate user and group quotas correctly based on the new param setting.

I did find reference to the "script," but it's not really a script, and it's only used for Tree Quota recalculation following a change as mentioned above.

Did this help?

dynamox · Answer

have you seen this: emc102246

Rainer_EMC · Answer

Is the cifs.useQuotaInFsStat server parameter set? doesnt matter - this param only governs NFS clients

Rainer_EMC · Answer

have you seen this: emc102246 Fixed in NAS Code 5.3.19 and above, or in 5.4.16 and above - which was ages agowhich DART release are you using ?

dynamox · Answer

i found it emc120377 ...yep ..it's only for tree quota issue.

Celerra

CIFS quota reporting file sizes/file count incorrectly

Was this post helpful?