[18944] in Athena Bugs
linux 8.4.20: fsid
daemon@ATHENA.MIT.EDU (Ken Raeburn)
Tue May 8 23:08:22 2001
Message-Id: <200105090308.XAA10128@all-in-one.mit.edu>
To: bugs@MIT.EDU
Cc: raeburn@MIT.EDU
Date: Tue, 08 May 2001 23:07:13 -0400
From: Ken Raeburn <raeburn@MIT.EDU>
System name: all-in-one.mit.edu
Type and version: i686 8.4.20
Display type: XFree86 3.3.6 SVGA
Shell: /bin/athena/tcsh
Window manager: unknown
What were you trying to do?
remap to all AFS cells I'm accessing with attach
% attach krb5tools@raeburn.org
attach: /afs/athena.mit.edu/user/r/a/raeburn/.raeburn.org/project/krb5tools attached to /mit/krb5tools for filesystem krb5tools@raeburn.org
% fs whichcell /mit/krb5tools/.
File /mit/krb5tools/. lives in cell 'raeburn.org'
...
wait for tokens to expire, then get new tickets
...
% fsid -a
...
fsid: krb5tools@raeburn.org mapped
...
% tokens
Tokens held by the Cache Manager:
User's (AFS ID 7882) tokens for afs@sipb.mit.edu [Expires May 9 03:51]
User's (AFS ID 7882) tokens for afs@dev.mit.edu [Expires May 9 03:51]
User's (AFS ID 7882) tokens for afs@athena.mit.edu [Expires May 9 03:51]
--End of list--
%
(The weird path is because Athena doesn't have /afs/raeburn.org and I
don't want to change the local cell to raeburn.org. But cell
raeburn.org is in my local CellServDB.)
What's wrong:
fsid doesn't seem to check the cell of the final component of
the path
What should have happened:
Both attach and fsid (really, afs_auth in liblocker) should do
an "fs whichcell" kind of operation on the final pathname.
Maybe it should also map to cells of intermediate pathnames if
needed to gain access to the final pathname.
That would, unfortunately, increase AFS traffic and slow down
fsid and attach. Perhaps attachtab could hold some extra
info, like the cell name(s), but that's not necessarily
correct either since mount points or symlinks could change.
(As a workaround, yes, I should be able to use
$FSID_EXTRA_CELLS.)
Please describe any relevant documentation references:
fsid(1), attach(1)
