[Gluster-devel] Question on quota-anon-fd-nfs.t

[Shyam]

On 10/20/2014 09:25 AM, Niels de Vos wrote:
> On Mon, Oct 20, 2014 at 01:40:01PM +0200, Emmanuel Dreyfus wrote:
>> Raghavendra G <raghavendra.hg at gmail.com> wrote:
>>> 2. whether gluster volume is exported through NFS. For eg., on my system,
>>> where gluster volume ptop is exported through NFS,:
>>
>> That works:
>>
>> # showmount -e
>> Exports list on localhost:
>> /patchy
>>
>> But:
>>
>> # mount -t nfs localhost:patchy /mnt2mount_nfs: rpcbind to nfs on
>> server: RPC: Program not registered
>>
>> And here is the probable cause: the IP/port for mountd and nfsd seems
>> kinked. I don't understand how it finds mountd but it does not find
>> nfsd.
>
> Note that Gluster/NFS only supports NFSv3. Maybe you need to pass
> something like '-o vers=3' to mount? The Linux mount.nfs fallsback to V3
> in case V4 fails, I would expect NetBSD to do so too. It seems that
> tests/nfs.rc in the master branch has the NetBSD specific options.
>
> Aha! tests/basic/quota-anon-fd-nfs.t does not use the helper function,
> but executes 'mount -t nfs ..'.
>
> I guess you know better than most of us how to fix that :)

This script should also do the sleep/wait trick like other NFS test 
scripts after the volume is started, to give gluster NFS a chance to 
start and register the NFS service. (this is basically the showmount 
only, but better to add it to the script)

Or IOW,
Add, EXPECT_WITHIN $NFS_EXPORT_TIMEOUT 1 is_nfs_export_available
Before, the mount, just to be sure.

Shyam