[14236] in Athena Bugs
Re: cache corruption problem on Sparc5's (?)
daemon@ATHENA.MIT.EDU (mbarker@MIT.EDU)
Mon Mar 4 17:02:20 1996
From: mbarker@MIT.EDU
Date: Mon, 4 Mar 96 17:02:09 -0500
To: eichin@MIT.EDU, Michael J Duff <mjduff@MIT.EDU>, yandros@MIT.EDU
Cc: Marc Horowitz <marc@MIT.EDU>, bugs@MIT.EDU, jweiss@MIT.EDU, miki@MIT.EDU
Hi. Our best knowledge of the problem (over here in dcns-dev) is that it
seems to occur on a heavily loaded machine. blocks in AFS space are left
out of order afterwards ("shuffled" is the way someone put it).
We've mostly seen the problem when doing large builds. This may simply be
a result of the fact that such a build uses a lot of file accesses and does
some minimal checking on what kind of results it has.
We have not had good luck on reproducing the problem regularly enough to show Transarc
what we are griping about. I believe we now have open problem reports with
them on the issue, though.
If you can show us a make or some set of actions that fairly often produces the
problem, we would be VERY grateful.
Thank you for using Athena
Mike Barker
