Re: bin/48774: quota tests fail with NFS root

[David Holland <dholland-bugs%netbsd.org@localhost>]

The following reply was made to PR bin/48774; it has been noted by GNATS.

From: David Holland <dholland-bugs%netbsd.org@localhost>
To: gnats-bugs%NetBSD.org@localhost
Cc: 
Subject: Re: bin/48774: quota tests fail with NFS root
Date: Thu, 1 May 2014 16:42:27 +0000

 On Wed, Apr 30, 2014 at 08:45:00PM +0000, martin%NetBSD.org@localhost wrote:
  > I run atf tests on a system with / on NFS (nfs server is netbsd, no quotas
  > enabled on the underlying filesystem). Most quota related tests fail,
  > see for example:
  > 
  > 
http://www.netbsd.org/~martin/evbarm-atf/6_atf.html#fs_ffs_t_getquota_get_be_1_both
  > 
  > The quota command outputs two ERANGE error messages before printing the
  > correct values for the rump hijacked file system image.
 
 The overt problem seems to be that it's getting the quotas for /
 rather than for the rumpity test volume. It's odd that it should get
 the correct values too; maybe the test is running quota -a and
 filtering it for the intended volume? I don't particularly know what
 the test does in detail, as jmmv wrote it.
 
 That said, if it queries your nfs root for quotas it should get either
 results or no results, not errors. Does the nfs server run rquotad? It
 is possible that wrong/stupid things happen on nfs servers that don't
 as this probably hasn't been tested very much. If the nfs server's
 rquotad is replying with nonsense it would probably be a good idea to
 find out what and why.
 
 If you run quota on the nfs root volume outside of the rumpity test
 stuff, what does it print?
 
 -- 
 David A. Holland
 dholland%netbsd.org@localhost