Re: Linux 2.6 NFS client read-ahead

[Trond Myklebust <trond.myklebust@fys.uio.no>]

On Wed, 2006-05-24 at 13:53 +0200, German San Agustin wrote:
> We have several linux 2.6.9 accesing to a netapp filer via nfs to
> update several thousands of files. The application only need to read a
> few blocks of the files when updating so we found out that disabling
> the read-ahead on the server improve greatly the performance by
> maintaining a clean cache and decreasing the number of access to disk.
> We have been trying to disable the read-ahead in the client as well to
> reduce the access to the server even further; but we couldn't find
>  where to do this in the 2.6 kernel family. Is it possible?, or it is
> simply that on 2.6 kernels it is not possible to tune the nfs client.

It is not possible, and in this case, it really shouldn't make much
difference to you either as the readahead code is adaptive: it should
automatically detect that you are doing short reads.

If you want to check that it is working, I suggest using tcpdump to
monitor the NFS traffic to that file, then comparing with a "strace"
dump.

Cheers,
  Trond

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