Disk quota exceeded, and unable to rm a file to get it back.

167 views
Skip to first unread message

Christopher Swenson

unread,
Jan 20, 2014, 6:23:40 PM1/20/14
to sage-...@googlegroups.com
This is actually kinda funny. I was downloading a copy of Wikipedia to do some analysis, forgetting that there is a 5GB quota.

So, it fails to download due to disk exceeded. But the file is still there with the partial download, and rm fail because of disk quota.

Whoops.

Inline image 1

--Christopher
image.png

William Stein

unread,
Jan 20, 2014, 6:34:41 PM1/20/14
to sage-...@googlegroups.com

On Jan 20, 2014 3:24 PM, "Christopher Swenson" <ch...@caswenson.com> wrote:

This is actually kinda funny. I was downloading a copy of Wikipedia to do some analysis, forgetting that there is a 5GB quota.

So, it fails to download due to disk exceeded. But the file is still there with the partial download, and rm fail because of disk quota.


Please insert a quarter to continue :-)

But seriously, send me your project id so I can increase the quota.

This is one of those counterintuitive aspects of ZFS.     Eventually there will be a snapshot deletion feature, which will be useful in this case. 

William
 

Whoops.

Inline image 1

--Christopher

--
You received this message because you are subscribed to the Google Groups "sage-cloud" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sage-cloud+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/sage-cloud/CACBmJOVn6hgYafkAycOoWAESHkuTpnqHFDe2TRMGndRNRRL8Kg%40mail.gmail.com.
For more options, visit https://groups.google.com/groups/opt_out.
image.png

Christopher Swenson

unread,
Jan 20, 2014, 7:09:15 PM1/20/14
to sage-...@googlegroups.com
Project f76bd5bb-7fea-4837-8a2e-7ef818e49978 -- I'll use a scaled down dataset for now on SMC to do my work. :)

I was reading about this "feature" of ZFS. Unfortunately, the standard remedies, truncating the file with cp /dev/null or cat /dev/null or dd if=/dev/zero or truncate or > don't seem to work either.

--Christopher


image.png

William Stein

unread,
Jan 20, 2014, 7:22:49 PM1/20/14
to sage-...@googlegroups.com
On Mon, Jan 20, 2014 at 4:09 PM, Christopher Swenson <ch...@caswenson.com> wrote:
Project f76bd5bb-7fea-4837-8a2e-7ef818e49978 -- I'll use a scaled down dataset for now on SMC to do my work. :)

You can also use /scratch if it's just temporary data...

Your quota should now be 20GB on that project. 
 

I was reading about this "feature" of ZFS. Unfortunately, the standard remedies, truncating the file with cp /dev/null or cat /dev/null or dd if=/dev/zero or truncate or > don't seem to work either.

With ZFS there is no going back... which has its pluses! 

I do have a *lot* of hard drive space, and a lot of room to add more.  The main issue is more about how to manage all the data, rather than having space for it.  E.g., importing a ZFS pool with 10,000 filesystems on it takes a while.  And backing everything up offsite properly (which I'm doing) presents interesting challenges too...  

William
 

For more options, visit https://groups.google.com/groups/opt_out.



--
William Stein
Professor of Mathematics
University of Washington
http://wstein.org
image.png

Christopher Swenson

unread,
Jan 20, 2014, 7:26:51 PM1/20/14
to sage-...@googlegroups.com
Thanks!

What's /scratch?

~$ ls /scratch
ls: cannot access /scratch: No such file or directory


image.png

William Stein

unread,
Jan 20, 2014, 7:28:38 PM1/20/14
to sage-...@googlegroups.com
On Mon, Jan 20, 2014 at 4:26 PM, Christopher Swenson <ch...@caswenson.com> wrote:
Thanks!

What's /scratch?

~$ ls /scratch
ls: cannot access /scratch: No such file or directory


Oops -- I never created scratch on the Google Compute Engine nodes, and your project is on one.  I'll add some /scratch disks now. 

William

 

For more options, visit https://groups.google.com/groups/opt_out.
image.png

William Stein

unread,
Jan 20, 2014, 7:41:36 PM1/20/14
to sage-...@googlegroups.com
On Mon, Jan 20, 2014 at 4:28 PM, William Stein <wst...@gmail.com> wrote:



On Mon, Jan 20, 2014 at 4:26 PM, Christopher Swenson <ch...@caswenson.com> wrote:
Thanks!

What's /scratch?

~$ ls /scratch
ls: cannot access /scratch: No such file or directory


Oops -- I never created scratch on the Google Compute Engine nodes, and your project is on one.  I'll add some /scratch disks now. 

William


OK, there is now a 64GB /scratch on every compute machine (google or not).  They are all compressed deduped ZFS filesystems, but with *no* snapshotting.   I'll make them bigger later when I add more disk capacity to the zpools...
image.png
Reply all
Reply to author
Forward
0 new messages