> Historically, if there any features to be tripped over in the nfs client
> or server, I'll be guaranteed to stumble on them. All looks good here.
> Steve
Actually, I know for a fact one aspect of knfsd that is completely broken.
The NFS reply cache (fs/nfsd/nfscache.c) doesn't work in the case of
replies where results are (supposed to be) cached. I have sent some mail
to the list (including a workaround patch which doesn't get at the heart
of the problem but which Works For Me(TM) :-) and to G. Allen Morris, the
current maintainer.
This doesn't seem to affect linux clients but is deadly for Solaris
clients. The bug only crops up if a (cacheable) reply from the linux
server is dropped by the network or the remote client, so it may not hit
users with unsaturated networks.
If anyone is interested in discussing this further, let me know. I have
two separate patches which 'fix' this, but I don't have a deep
understanding of the code, so I can't say for sure that either is correct.
I will definitely NOT be running the as-is knfsd (i.e. without my patch)
on my server because it wreaks havoc on Solaris clients.
David
-- /==============================\ | David Mansfield | | david@cobite.com | \==============================/
- 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/