quota -f doubling users' quota usage

[Jack Neely]

On Thu, Nov 05, 2009 at 03:36:24PM +1100, Bron Gondwana wrote:
> On Wed, Nov 04, 2009 at 11:17:07AM -0500, Jack Neely wrote:
> > Folks,
> > 
> > I'm in the midst of upgrading our cyrus imap servers.  We've hit a snag
> > that our testing did not reveal.  The quota -f utility is doubling all
> > of our users' actual usage.  This seems to be reported several times on
> > various places on the web.  Running quota -f a second time seems to
> > correct the issue.  However, I wanted to track down the actual issue/bug
> > here.
> > 
> > We are moving from RHEL 3 i386 with Cyrus version 2.2.12 to RHEL 5
> > x86_64 machines running cyrus version 2.3.14.  We tar up the cyrus data
> > store and move it to the new server, rebuild the DB from a dump from the
> > old server, run reconstructs, and finally run quota -f.  We are getting
> > output below.
> > 
> > user.cechafin.RESTORE.Drafts: quota root user.cechafin.RESTORE --> user.cechafin 
> > user.cechafin.RESTORE.Sent: quota root user.cechafin.RESTORE --> user.cechafin  
> > user.cechafin.RESTORE.Sent.2006: quota root user.cechafin.RESTORE --> user.cechafi
> > user.cechafin.RESTORE.Sent.2007: quota root user.cechafin.RESTORE --> user.cechafin
> > user.cechafin: usage was 33900479, now 33914622
> > user.mmorrow: usage was 152034181, now 105570337
> > user.vrtownse: usage was 9382963, now 9385921
> > user.wwang: usage was 203145147, now 203333738
> > user.aaardest: usage was 31204975, now 62409950
> > user.aaaskar: usage was 44789845, now 89579690
> > user.aaatiq: usage was 29105529, now 58211058
> > 
> > What's happening?
> 
> I _suspect_ that the changing of the quota roots above is causing a second pass through
> a loop somewhere.  The code is pretty, um, loopy.
> 
> You say a second quota -f run fixes it.  I suspect that doing a second quota -f run
> is the correct response.  Yes there's a bug in there somewhere, but there's a pretty
> good workaround available :)
> 
> Bron ( but put in a bugzilla entry if there isn't one already, and we'll notice it
>        again one day )

Bugzilla is my friend.  :-)

https://bugzilla.andrew.cmu.edu/show_bug.cgi?id=3179

A workaround is fairly easy, but I like to stomp on bugs.

Thanks!

Jack

-- 
Jack Neely <jjneely at ncsu.edu>
Linux Czar, OIT Campus Linux Services
Office of Information Technology, NC State University
GPG Fingerprint: 1917 5AC1 E828 9337 7AA4  EA6B 213B 765F 3B6A 5B89