Hi Steve-
There is an important case in mount.nfs where EACCES is temporary:
The kernel's rpcbind client returns EACCES if the remote rpcbind
server replies that the requested service is not registered.
It would be easy enough for the kernel's mount client to remap that
error code into something unique for the mount system call.
--
Chuck Lever
chuck[dot]lever[at]oracle[dot]com
Sorry for the delayed response... I spent most of last week
off line due to Red Hat moving to a new building... fun, fun, fun! :-\
Chuck Lever wrote:
> Hi Steve-
>
> There is an important case in mount.nfs where EACCES is temporary: The
> kernel's rpcbind client returns EACCES if the remote rpcbind server
> replies that the requested service is not registered.
I guess I see the reasoning behind this... a server could be coming
up so the mount should hang around waiting for that...
>
> It would be easy enough for the kernel's mount client to remap that
> error code into something unique for the mount system call.
I would think ENOENT would be a better mapping for RPC_PROG_UNAVAIL.
But that still leaves the possible problem of the mount not waiting
around for servers to come up... Or do we even have to worry about
mounts failing because a server is not up yet? I'm beginning to
think not...
steved.