Re: rm -r sometimes produces errors under NFS

[Paul Eggert]

Jim Meyering <address@hidden> writes:

> Paul Eggert <address@hidden> wrote:
>
>> Jim Meyering <address@hidden> writes:
>>
>>> When an NFS client sees a successful unlink, it is reasonable to
>>> expect a client-side rewinddir/readdir sequence *not* to produce
>>> the just-unlinked name.
>>
>> I agree, but won't this hurt rewinddir performance?  After all, one of
>
> Won't *what* hurt rewinddir performance?

I don't know the NFS details, but my assumption is that an obvious fix
for this bug would have the NFS client invalidate its cache of
directory entries after an unlink/rewinddir sequence.  Hence the
client would have to go back to the server on the next readdir (or
maybe stat), and that's another server round trip we'd like to avoid.
There might be better fixes but they'd require more memory on the
client.

(This is just a guess of course!  I haven't measured it.
I don't use NFS under Linux.)