https://bugzilla.novell.com/show_bug.cgi?id=880599
https://bugzilla.novell.com/show_bug.cgi?id=880599#c2
Neil Brown changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |nfbrown@suse.com
AssignedTo|nfbrown@suse.com |jack@suse.com
--- Comment #2 from Neil Brown 2014-06-05 01:38:39 UTC ---
Re-assigning to Jack as this seems to be an ext[34]/quotas problem more than an
NFS problem.
With NFSv3, the client doesn't find out about quota problems until the data is
actually sent to the server, and that can be delayed quite a while as writes
are cached on the client and sent lazily.
If you were to mount with "-o sync" you would get quota errors immediately, but
performance would be terrible, so that isn't a good idea.
If you reduce /proc/sys/vm/dirty_bytes to a few hundred meg, that would limit
the overage to a few hundred meg without hurting performance too much. However
that may still be more overage than you would like. How much memory does your
client have?
I cannot really see why NFSv4 would make a difference. It does have some
knowledge of quotas, were NFSv3 has none, but the Linux code doesn't appear to
use that at all.
I might try some experiments myself to see what happens when you exceed quotas
via NFS. However I think we need to start by resolving the ext3/4 issue as
that seems to be the main problem.
--
Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.