Re: nfs and amd
Linus Torvalds (torvalds@transmeta.com)
17 Apr 1997 03:02:28 GMT
In article <m0wHgU0-000HRiC@monad.swb.de>,
Olaf Kirch <okir@monad.swb.de> wrote:
>To: linux-kernel@vger.rutgers.edu
>Subject: Re: nfs and amd
>X-Newsreader: TIN [UNIX 1.3 950515BETA PL0]
>
>Michael Schulz wrote:
>: But then after amd unmounts the cdrom automaticaly and the nfsd
>: tries to access it again, i get the following error:
>:
>: Eimer nfsd[389]: impossible hash_path[0] value: 08060e00 e4
>: 9a0508000000000000000000000000004a4214067acecb31da7800
>
>This does not have anything to do with nfsd caching CDROM data (which
>it doesn't); it is really a bad NFS file handle. I have no idea
>what may be the cause of corruption; either nfsd or amd might be the one
>to blame. I am not familiar with the way amd lays out its file handles,
>but the above could easily be an amd handle being passed to nfsd by
>accident.
>
>Please get the latest nfsd from ftp.mathematik.th-darmstadt.de in
>/pub/linux/okir, run it with debugging enabled and send me the
>(trimmed) output.
Note that amd is fundamentally unstable, which is why Linux now has an
in-kernel autofs like Solaris does. It's only seen about a month of
testing, but I can already say with some certainly that it is about 10
times as stable as amd ever was (or is ever likely to be). At least for
us,
Linus