Mailing List Archive

David Lee <t.d.lee@durham.ac.uk> writes:
>
> If what follows is in the manual or is a FAQ, then just point me in the
> direction...
>
> Executive summary: Is there an easy way (variant of "quota report"
> perhaps?) to discover which users are over-quota?

[& from later in your posting...]

> Under Solaris etc., the output from its "repquota" command was easily
> eyeball-able (I know, no such word!) for such exceptions: "+" and expiry
> times.
>
> Is there a variant of NetApp's "quota report" that either restricts output
> to over-quota users, or has a field indicating over-quota, or similar?

Not that I know of.

> Or is the NetApp convention simply to cobble together a bit of awk/perl to
> extract those lines (from the total 20K) in which "Used" exceeds "Limit"?

I would use an awk or Perl script as you suggest. It doesn't have to be
more than a one-liner:

rsh [filer] quota report | awk '($1=="user") && (($5>$6) || ($7>$8))'

> Detail:
>
> We are beginning to migrate our home directories (15K-20K of them) from
> Solaris to NetApp. Quotas are a vital part of the control of these
> directories.

If you already have Solaris ufs quotas, why don't you use them to set
the NetApp ones, at least as a guideline? Usage won't be quite the same:
it goes up because the NetApp 4K allocation unit is larger than the 1K
fragment size on Solaris, and down because space occupied by directories
isn't accounted, but a 10% increase (say) would cover most cases.

> Under Solaris (trad. UNIX), each user has an explicit quota. While this
> is possible under NetApp, it seems far more sensible to use default quotas
> where possible (and the documentation encourages this).

It's true that using default quotas like that saves some overhead, but
nowadays it isn't all that significant. I use a default user quota of zero,
to be applied to uids that shouldn't have files in the home directory qtree
at all, and to cancelled users (we have a delay between cancelling a user
and removing their files, so that un-cancellation is easy). All the "real"
users have an explicit quota set in /etc/quotas, even though over 90% of
them have the same limits. Still, it sounds as though you have 3-4 times
as many users, so YMMV.

> We have set up some reasonable default quotas, and are using rsync
> (thanks, folk!) to get the user files prepared in the future NetApp
> location: transferred (from 20+ volumes across three Solaris fileservers)
> and kept in sync.

Ugh... I remember having to transfer the contents of 8 home partitions
spread across (guess!) three SunOS 4.1.3 servers to an FAServer 450
back in August 1995. Welcome to the club...

> Naturally a few of our users have more data than our NetApp default
> quotas. What we simply want to do is identify these exceptions so we can
> apply explicit quotas. We will, if possible, adjust the defaults so that
> the number of exceptions, from our 20K user-base, is reasonably small.

Hmmm... you do realise that this is giving to the rich (or the extravagant,
anyway), at the expense of the poor (or frugal)? :-)

Chris Thompson University of Cambridge Computing Service,
Email: cet1@ucs.cam.ac.uk New Museums Site, Cambridge CB2 3QH,
Phone: +44 1223 334715 United Kingdom.