[12334] in Athena Bugs
Re: decmips 7.7E: fs
daemon@ATHENA.MIT.EDU (cfields@MIT.EDU)
Fri Aug 5 23:54:12 1994
From: cfields@MIT.EDU
Date: Fri, 5 Aug 1994 23:53:41 +0500
To: jweiss@MIT.EDU
Cc: bugs@MIT.EDU, testers@MIT.EDU, bug-afs@MIT.EDU
> 1) fs listcells produces corrupt output:
I failed to reproduce this problem on my DECstation. I ran the fs in
release 7.7 as well as the one from 7.6, and they produced identical
outputs.
Both outputs were correct, except for the first cell listed:
Cell on hosts lpd1.ucop.edu 120.86.52.18 128.48.141.9 msds.ucop.edu.
A 7.7 Sun returns:
Cell ucop.edu on hosts lpd1.ucop.edu msds.ucop.edu 128.48.141.9.
> 2) fs newcell of an existing cell does not do anything. I
> managed to fs newcell the lees cell with the 7.6 fs; it was also
> producing less corrupt (but not perfect) output to fs listcells.
Given this information, and what I just did, it's not clear to me
that this is a 7.7 bug. It sounds like strange things may have been
happening on your workstation, and without knowing more detail of
what you did, it's entirely possible that the 7.6 fs succeeded where
the 7.7 fs failed on the newcell simply because it was lucky.
I would like you to try the following things:
1. reproduce the output error with the 7.7 fs on your workstation
2. see if the 7.6 fs also produces bad output once the 7.7
fs is producing bad output
if 7.6 does not produce bad output at the same time that
7.7 does, you're done
3. if you can't immediately reproduce the error, try to
remember what you were doing with AFS befor the error
occurred, and try to reproduce it in light of that
information. If reproduction involves using 7.7 fs commands,
see if you can also reproduce it using 7.6 fs commands. You
may find it useful to reboot the workstation to see
if that clears the corrupted state before the next cycle
of reproduction.
4. if you can reproduce with 7.7 fs, but not 7.6 fs, we're set
5. if you can reproduce it with both, you need to find a 7.6
workstation and try to reproduce the problem there
6. if you can't reproduce it at all, ummm....
Craig
