[OpenAFS] Windows 7 x64: \\afs\cs.uwm.edu refers to a location that is unavailable

Jeffrey Altman jaltman@secure-endpoints.com
Thu, 28 Apr 2011 15:39:23 -0400


This is an OpenPGP/MIME signed message (RFC 2440 and 3156)
--------------enig6B04785158FC952C76DB93A4
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

On 4/28/2011 1:20 PM, John Tang Boyland wrote:
> Here is the problem I alluded to earlier:
>=20
> I have a student who bought a new laptop with Windows Version 6.1.7600 =
and=20
> installed NIM and OpenAFS 1.5.9904 and things ALMOST work.
>=20
> They can get tokens and can go to \\afs\openafs.org
> and \\afs\cs.wisc.edu but not \\afs\cs.uwm.edu
> When they try, they get the message:
> "\\afs\cs.uwm.edu refers to a location that is unavailable.  It could b=
e
>  on a hard drive on this computer, or on a network.  Check..."

this means that the evaluation of the mount point "\\afs\all\cs.uwm.edu"
is incorrect.  This could be a bug introduced in 1.5.9904 which is fixed
in the source tree waiting for 1.5.9905 (1.6.0pre5) to be released.

"fs lsm \\afs\all\cs.uwm.edu" will report the mount point info.
If its wrong, "fs rmm \\afs\all\cs.uwm.edu" and "fs mkm
\\afs\all\cs.uwm.edu <correct-target>"

Or it could be something else....

> They can run vos commands: "vos examine root.cell" works fine.

vos commands do not use the cache manager or the mount point definition
in the freelance root volume.

> If they run "fs checks", there is a long pause, and then it says that
> "filip.cs.uwm.edu" is unavailable, even though it pings fine from this
> machine and the server is accessible from elsewhere.
> (filip is the RW location of root.cell, but it is also available on two=

> other servers: solomons and jeremiah)  The main DB server is solomons.

"fs memdump" will produce a file %windir%\temp\afsd_alloc.log that dumps
the contents of a large number of cache manager objects.  At the bottom
of the list is the list of servers and their status.

A cache manager trace for the operations can be obtained with

"fs trace -reset -on"
"fs checkservers"
"fs trace -dump -off"

which produces %windir%\temp\afsd.log and records the operations of all
of the threads in the userland cache manager.

> NET VIEW \\AFS shows
>=20
> Share name =20
> ...
> cs.uwm.edu   Disk  <UNC>    AFS MountPoint to #cs.uwm.edu:root.cell.
> cs.wisc.edu  Disk           AFS MountPoint to #cs.wisc.edu:root.cell
> openafs.org  Disk           AFS MountPoint to #openafs.org:root.cell
>=20
> NB: The "period" at the end of the cs.uwm.edu entry makes me worried
> now.  Is that a possible problem?  And why are ONLY the cells that
> cannot be reached marked as UNC?  But if the period was the error, why
> would it say that filip was unavailable?

The period is the problem.  That is a bug that has been fixed.  You can
correct it locally by following the instructions above.

The server being offline is a separate issue entirely.

Jeffrey Altman


--------------enig6B04785158FC952C76DB93A4
Content-Type: application/pgp-signature; name="signature.asc"
Content-Description: OpenPGP digital signature
Content-Disposition: attachment; filename="signature.asc"

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (MingW32)

iQEcBAEBAgAGBQJNucJsAAoJENxm1CNJffh4qlYH/3iJIE5lV2JeXHj4I0Co1Lzn
p8cGA1ddVmwQ8rBpgOOBddsw/9vgmtPRvIkDy5PUR4oftuJ2FhLSNxCYrpodIRGr
XDI7G/xi+g9dB7TVKTDWC/xVHMj7hMLCANrDmlOR2yP67KhZOTKPjQxzsqYRTbjE
xkQDtXkOEprUWrEJGdulS9o6kqF0CDdjBrYMif52fk9GD92/XODZOGQ1uoOfj+l2
rE01qW5oEf21LSHfQHM/qybzbhAHrx3wcFNozT6f1/vpofaMSnYhBwtqOtQPOj09
lk78ga+cMgcBr+kv4W0opDpR6MxVWcd9lPbBHM/Vf9W8O/kBFN4dbQk8InWiGzY=
=7R6c
-----END PGP SIGNATURE-----

--------------enig6B04785158FC952C76DB93A4--