Re: NFS in 2.2.1-ac3 - cache not synched ?

Thierry Danis (danis@sagem.fr)
Thu, 4 Feb 1999 11:46:25 +0100 (MET)

>
> > It writes a 1,8 Gb file, rereads it, then unlinks it and exits (iozone
> > test program).
>
> Ok
>
> > Once the file has been unlinked, I verify that it has disappeared on
> > both ends (serveur and client). But a 'df' indicates that the space
> > is still used (ie, there is still 1,8 Gb used on the serveur partition).
>
> That sounds like the handle is being cached. Unfsd also has this feature
> but releases a handle in 30 seconds. Im not sure what knfsd does with
> timeouts.
>
> How long did you leave it ?
>

A long time !

> Alan
>

In fact, as long as I do not create any new file (or umount/mount
the NFS partition), the handle seems to be cached.
If I create a file from another NFS client, it takes a long
time for the creation to happen (the server is freeing the
space before doing some creation).

On the other hand, issuing a 'rm' on the client is almost
instantaneous, even for the 1,8G file (since the blocks
are actually not freed). But it takes a long time for
the next access to begin. And since some tools tests
for the available space before creating files, I am not
sure it is advisable to delay the actual freeing up to
the next operation.

A+

-- 
	Thierry Danis


-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/