Unsolved
This post is more than 5 years old
16 Posts
0
840
Quota over NFS4 is sometimes reported wrong using df
Hi,
QUESTION:
------------------------------------
We have a cron job running which regularly checks users' and groups' disk usage and quota. We get the values with df.
Occasionally it happens that the values for one user or group are complete nonsense. The next time the check runs everything is fine again. Look at this log excerpt:
21.02.2014, 19:02:08 Schicke Warnung an user jongleur: 98% (58 GB / 59 GB)
21.02.2014, 19:32:06 Schicke Entwarnung an user jongleur: 1% (0 GB / 1 GB)
21.02.2014, 21:02:05 Schicke Warnung an user fschyma: 94% (17 GB / 18 GB)
21.02.2014, 21:32:07 Schicke Entwarnung an user fschyma: 0% (0 GB / 1 GB)
22.02.2014, 04:02:08 Schicke Entwarnung an user loock: 77% (378.4 GB / 489 GB)
22.02.2014, 04:32:10 Schicke Warnung an user loock: 90% (1.8 GB / 2 GB)
As you can see the check runs every 30 minutes.
How can this be possible?
BUSINESS IMPACT:
------------------------------------
Users and groups receive wrong alert or recovery emails about their used disk space.
Peter_Sero
1.2K Posts
0
February 25th, 2014 00:00
Wirklich 'ne "schicke Warnung"...
Are you running the df on an NFS client,
and the quotas are directory-based and "container" enabled?
I have seen wrong df reports in such a situation, on 6.5.5.x.
Sporadically the size and usage numbers
of NFS share X are identically reported for NFS share Y.
Pretty strange, but as we use df over NFS only for
quick online checks, it hasn't mattered much to us.
Cheers
-- Peter