[61] in 6.033-lab
Re: final things.
daemon@ATHENA.MIT.EDU (Frans Kaashoek)
Mon May 12 07:14:31 1997
To: "Kevin 'Frostbyte' McCormick" <fbyte@sub-zero.mit.edu>
Cc: 6.033-lab@MIT.EDU
In-Reply-To: Your message of Mon, 12 May 1997 05:13:23 -0400.
Date: Mon, 12 May 1997 07:14:15 -0400
From: Frans Kaashoek <kaashoek@amsterdam.lcs.mit.edu>
In message <Pine.LNX.3.93.970512050921.4043A-100000@sub-zero.mit.edu>, fbyte@su
b-zero.mit.edu writes:
>
>Okay, so I have this filesystem, and it's all nice and finished except for
>one thing: (this is on Linux)
>
>If, for example, I run the NFS loopback server on a filesystem, ls /sfs or
>whatever, then kill the server, make a clean filesystem, rerun the server,
>then ls /sfs, all the old files are still there. The kernel caches NFS
>data and for the life of me I can't figure out what I have to do in the
>NFS server to make the kernel realize that its cached data is invalid.
>The way I've been doing it for a while now is to run my sfscd, then the
>pristine sfscd, then mine again, which invalidates the caches. But, of
>course, this is unacceptable. Any ideas?
>
On Linux I believe there is a flag to mount that allows you to specify that
the kernel should not cache attributes. On operating systems that do no
have such a flag (e.g., OpenBSD), you have to flush the attribute cache in
some hacked-up way.
Frans
