[1948] in SIPB_Linux_Development
Re: AFS cache corruption under 2.0.32
daemon@ATHENA.MIT.EDU (Derek Atkins)
Tue Dec 2 15:21:01 1997
To: Greg Hudson <ghudson@MIT.EDU>
Cc: "Theodore Y. Ts'o" <tytso@MIT.EDU>, Emil Sit <sit@MIT.EDU>,
linux-dev@MIT.EDU, linux-afs-bugs@MIT.EDU
From: Derek Atkins <warlord@MIT.EDU>
Date: 02 Dec 1997 15:15:25 -0500
In-Reply-To: Greg Hudson's message of Tue, 02 Dec 1997 14:03:39 EST
So, perhaps we should have the shutdown code write a file and when AFS
starts clear the cache if that file doesn't exist and remove the file
if it does.
This way, the cache will remain if AFS cleanly shuts down, and the
cache will be blown away if the if it doesn't.
I'm not sure if this should happen in the AFS code or in the Athena
code.
Comments?
-derek
Greg Hudson <ghudson@MIT.EDU> writes:
>
> > If the AFS cache corruption only occurs when the machine is
> > uncleanly shutdown, I'm not sure we can really blame the AFS
> > code..... does any AFS port handle that condition cleanly?
>
> No other AFS port is known to have such dramatic or frequent lossage
> when a machine experiences an unclean shutdown. It's common practice
> to use stop-A and "boot" to reboot cluster suns, for instance.
>
> However, you're probably right that it's probably not AFS's problem.
> I suspect that it's because ext2fs doesn't make even a vague attempt
> at staged writes like FFS does.
--
Derek Atkins, SB '93 MIT EE, SM '95 MIT Media Laboratory
Member, MIT Student Information Processing Board (SIPB)
URL: http://web.mit.edu/warlord/ PP-ASEL N1NWH
warlord@MIT.EDU PGP key available
