Re: [RFC][PATCH 2/4] sunrpc: Use utsnamespaces

[Matt Helsley <matthltc@us.ibm.com>]

On Tue, 2009-01-06 at 15:44 -0500, Trond Myklebust wrote:
> On Tue, 2009-01-06 at 14:02 -0600, Serge E. Hallyn wrote:
> > Quoting Matt Helsley (matthltc@us.ibm.com):
> > > We can often specify the UTS namespace to use when starting an RPC client.
> > > However sometimes no UTS namespace is available (specifically during system
> > > shutdown as the last NFS mount in a container is unmounted) so fall
> > > back to the initial UTS namespace.
> > 
> > So what happens if we take this patch and do nothing else?
> > 
> > The only potential problem situation will be rpc requests
> > made on behalf of a container in which the last task has
> > exited, right?  So let's say a container did an nfs mount
> > and then exits, causing an nfs umount request.
> > 
> > That umount request will now be sent with the wrong nodename.
> > Does that actually cause problems, will the server use the
> > nodename to try and determine the client sending the request?
> 
> The NFSv2/v3 umount rpc call will be sent by the 'umount' program from
> userspace, not the kernel. The problem here is that because lazy mounts

Ahh, that's news to me. I thought userspace originated the umount but
the kernel constructed and sent the corresponding RPC call.

> exist, the lifetime of the RPC client may be longer than that of the
> container. In addition, it may be shared among more than 1 container,
> because superblocks can be shared.

Right.

> One thing you need to be aware of here is that inode dirty data
> writebacks may be initiated by completely different processes than the
> one that dirtied the inode.
> IOW: Aside from being extremely ugly, approaches like [PATCH 4/4] which
> rely on being able to determine the container-specific node name at RPC
> generation time are therefore going to return incorrect values.

Yes, I was aware that the inode might be dirtied by another container. I
was thinking that, at least in the case of NFS, it makes sense to report
the node name of the container that did the original mount. Of course
this doesn't address the general RPC client case and, like patch 3, it
makes the superblock solution rather NFS-specific. That brings me to a
basic question: Are there any RPC clients in the kernel that do not
operate on behalf of NFS?

Thanks!

Cheers,
	-Matt Helsley