[11170] in Athena Bugs
vax 7.4H: "Unknown error" when saving if over quota
daemon@ATHENA.MIT.EDU (probe@MIT.EDU)
Wed Oct 6 17:04:52 1993
From: probe@MIT.EDU
Date: Wed, 6 Oct 1993 17:04:31 -0400
To: tmgreen@MIT.EDU, bugs@MIT.EDU
Cc:
>From tmgreen@Athena.MIT.EDU Fri Aug 27 09:01:38 1993
>Received: by cascade.MIT.EDU (5.57/4.7) id AA27106; Fri, 27 Aug 93 09:01:37 -0>400
>Received: from SUM4.MIT.EDU by Athena.MIT.EDU with SMTP
> id AA19568; Fri, 27 Aug 93 09:01:30 EDT
>From: tmgreen@Athena.MIT.EDU
>Received: by sum4.MIT.EDU (5.61/4.7) id AA08718; Fri, 27 Aug 93 09:01:21 -0400>
>Message-Id: <9308271301.AA08718@sum4.MIT.EDU>
>To: bugs@Athena.MIT.EDU
>Subject: vax 7.4H: "Unknown error" when saving if over quota
>Date: Fri, 27 Aug 93 09:01:18 EDT
>
>System name: sum4
>Type and version: CVAXSTAR 7.4H
>Display type: SM
>
>What were you trying to do?
>
>Save a file in emacs
>
>What's wrong:
>
>I tried to save a file from emacs and I got the error:
>
>"Opening output file: undocumented error code, /afs/athena/....(myfile)"
>
>Not only did it fail to write the file, but it ERASED it from disk.
>I then tried to write (\C-X-\C-W) the buffer as another file name and
>got the same error.
>
>I tried to do the same with another buffer and the file got ERASED,
>too. This buffer was my ENTIRE THESIS, which was pretty distressing.
>Then, as a last resort, I tried writing the file into another locker
>that I had attached and it worked.
>
>I finally figured out by logging in to another workstation that I was
>slightly over my disk quota.
>
>(a) why couldn't it tell me this?
>(b) why did it delete my instead of just refuse to save it?
When we moved to a newer version of AFS, there was an incompatible change
between old and new software releases. The servers are running AFS 3.2 and
all of the Athena clients EXCEPT the VAX are running AFS 3.2. The VAX, for
political reasons, is still running AFS 3.1, which means that the error
codes are not being mapped properly. Since the VAXstations are being phased
out and no development time is being spent on the VAX, it is highly unlikely
this bug will be fixed.
Essentially, the fix is to replace the /vmunix kernel on all the workstations,
but because there are so few VAXen around, it was felt that the new kernel
would receive inadequate testing, and therefore, it was never released.
I apologize for the loss of work. For the past several months, this error
has occurred, and the consultants were informed as to its meaning.
I believe that "emacs" probably did tell you that it was unable to save the
file, but the problem is that for it to have discovered the problem, it had
to make the attempt. Be very wary whenever you get an error saving, as it
may mean that it has already attempted the operation (in fact, that is almost
always the case with any utility). It was the server that reported the error
during the write attempt - the client doesn't know what errors the server may
report or what it might permit, so it has to try.
-Richard Basch
