On Tue, 13 Jan 2009, David Lee wrote:
On Tue, 13 Jan 2009, Darren Miller wrote:
[...] Me too. We contacted NetApp and got the fix which allowed us to turn quotas back on. However, quotas are now "leaking", which means quota usage is actually higher than the actual disk usage, causing many of our users to hit their quota limit. It only affects CIFS, NFS quotas are fine.
Ugh. Thanks for the heads-up.
Would that fix happen to be 'wafl_enable_allocation_size 0'? We applied that fix and restarted quotas earlier today, and are holding our breath...
Warning: side-effect: CPU gobbling by "quota resize".
Our operational procedures includes handling Servicedesk (Helpdesk) requests from users for increased quotas. Our staff use a UNIX script which (1) edits the Netapp "quotas" file and (2) invokes "quota resize" on the relevant volume. This has worked happily for years. But since doing that "wafl_..." workaround yesterday, we've had bursts of very poor response from the filers. Closer examination, and a controlled test, show that these have coincided with invoking "quota resize".
It looks as though that "wafl_..." workaround and "quota resize" are most unhappy partners.
I hope that the proper fix to the quota-crash problem won't introduce this performance-hit as a side-effect...
Is there a NetApp person on this list taking an overview of this particular issue, and collating information?