NFS mount fails CentOS5 on FC16 host

[Bill Davidsen]

Veli-Pekka Kestilä wrote:
> On 10.5.2012 0.40, Bill Davidsen wrote:
>> Ed Greshko wrote:
>>> On 05/09/2012 10:26 PM, Bill Davidsen wrote:
>>>> I get a message "rpc.idmapd appears not to be running" when it is running,
>>>> and I'm
>>>>
>>>> mount is "rw,soft,intr -t nfs4" and the idmap process is running on host and
>>>> client, remounts from other clients work, etc, etc.
>>> It isn't clear to me from the above what system is the Client and what system
>>> is the
>>> Server and on which system you're getting the error message.
>>>
>> Sorry, I intended the subject to make that clear, the client is CentOS-5 the
>> server FC16. The message appears on the CentOS5 client. The same server and
>> data works on clients running FC9, FC10, FC13, RHEL-6, and I believe (ie. I'm
>> told but haven't personally tested) OpenBSD works as well.
>>
>> This is the only machine which stopped working when the NFS server was
>> updated, and of course it's the internal web server. :-(
>>
> I had lot of problems with NFSv4 and CentOS 5.8, you should check that you have
> latest kernel on 5.8 as that solved some mounting problems. As for the error
> messages telling rpc.idmapd not running you can safely ignore. The error comes
> from the mount.nfs4 trying to read file in var which is not in there.
>
> If you run strace mount.nfs4 server:/ /dir you will see that the command tries
> to access /var/lock/subsys/rpcidmapd
> access("/var/lock/subsys/rpcidmapd", F_OK) = -1 ENOENT (No such file or directory)
> This doesn't work as the file is actually named /var/lock/subsys/rpc.idmapd
> As how you get rid of the error before RedHat fixes the actual source (just make
> the soft link between rpcidmapd and rpc.idmapd)
>
Adding the link solved the problem, although it's a kludge for sure!

> Now to the real error, I had the mounting problems from CentOS 5.8 to CentOS 5.8
> with CLIENT running with kernel 2.6.18-308.1.1.el5 mounting didn't work. As I
> had the older kernel around as this was upgrade from older 5.x release I tried
> with it and the things worked without a problem. RedHat fixed the problem for me
> atleast in kernel 2.6.18-308.4.1.el5.
>
And there, too, the most recent kernel upgrade made it work (although the 
warning is still there without the link).

> Hope this helps,
> -vpk

Looks right, the spurious warning issue is avoided for the moment, and the 
failure of the mount is fixed (and hopefully added to regression testing and 
will stay that way). Thanks for the work-around on the warning, I had updated 
the kernel by the time I saw your message, since I was watching closely for updates.


-- 
Bill Davidsen <davidsen at tmr.com>
   "We have more to fear from the bungling of the incompetent than from
the machinations of the wicked."  - from Slashdot