Setting quotas on end points.

25 views
Skip to first unread message

Toby Archer

unread,
Jul 9, 2020, 10:27:29 AM7/9/20
to Admin Discuss
We have a Spectra Logic Black Pearl tape archive that we setting up to archive data across the South Dakota Board of Regents. Researchers pay a fee for a certain amount of data to be allocated to them. But the Black Pearl has no way of specifying how much data one person can use, it depends on the software that we use to control it to do that, which is Globus in our case. However, I don't see any likewise facility in Globus either. Does anyone know if Globus has a way of setting quotas/limits on the amount of data a user or endpoint can use?

~TA

Greg Nawrocki

unread,
Jul 9, 2020, 11:49:03 AM7/9/20
to admin-...@globus.org, Toby Archer
Hi Toby,

Sorry, Globus does not have any functionality to control per user storage quotas.

— Greg

Greg Nawrocki
University of Chicago
Globus
401 N Michigan Ave. - 9th Floor
Chicago, IL 60611

David Carver

unread,
Jul 9, 2020, 1:41:28 PM7/9/20
to Admin Discuss, sand...@gmail.com
Disk and tape space are not free and users hitting quota is one of our biggest issues.  Is there any plans to add a plug to check the quotas for the different file systems before writing the data to the filesystem.  Not only do users open tickets about transfers not working, but with a filesystem like lustre on a supercomputer where you have 26  OSTs nodes it can cause the filesystem to hang or crash, affecting hundreds of users.  Currently, our solution is to disable the userid and contact the user.  Even reporting an error message to the user that they may have hit a "quota limit" would help. 

Greg Nawrocki

unread,
Jul 9, 2020, 4:06:16 PM7/9/20
to admin-...@globus.org, Toby Archer, david....@gmail.com
At the moment we do not have plans to add a feature that would check the quotas on filesystems before writing data. However, there are two features related to quota errors on our radar:
We have been asked for supporting a flag that has the task fail on quota errors, and a notification is sent from the transfer service to the user on failures. When this option is chosen by the user it will override the default behavior where the task will keep retrying. This is being worked on now.
Another option that has been discussed is to pause the task and notify the user on the quota error. While that is in the backlog, it has not been prioritized.

Regarding indication that a quota limit may have been hit during a transfer, the transfer task will display a fault event in the event log with an indication of "System error in write: No space left on device" or "A system call failed: No space left on device”.

— Greg

Greg Nawrocki
University of Chicago
Globus
401 N Michigan Ave. - 9th Floor
Chicago, IL 60611

e: gr...@globus.org

Reply all
Reply to author
Forward
0 new messages