Date: Fri, 15 Mar 2013 09:53:21 -0400 (EDT) From: Rick Macklem <rmacklem@uoguelph.ca> To: Andre Oppermann <andre@freebsd.org> Cc: freebsd-current@freebsd.org Subject: Re: NewNFS vs. oldNFS for 10.0? Message-ID: <2090268844.3936361.1363355601938.JavaMail.root@erie.cs.uoguelph.ca> In-Reply-To: <514324E8.30209@freebsd.org>
next in thread | previous in thread | raw e-mail | index | archive | help
Andre Oppermann wrote:
> Hi Rick, all,
>
> is there a plan to decide for one NFS implementation for FreeBSD 10.0,
> or to keep both around indefinately?
>
> I'm talking about:
> oldNFS in sys/{nfs, nfsclient, nfsserver} NFSv2+NFSv3
> newNFS in sys/fs/{nfs, nfsclient, nfsserver} NFSv2+NFSv3+NFSv4
>
> NewNFS supports newer NFS standards and seems to have proven itself in
> some quite heavy traffic environments.
>
> Is there any reason to keep oldNFS around other than nostalgic?
>
I was planning on asking the "collective" that question at some point.
I think the newNFS works well enough that it can become the only NFS
in the kernel for 10.0, but I don't see that as my decision.
I am only aware of 2 possible regressions of the new one w.r.t. the
old one:
- A performance issue for the new server when under heavy NFS over TCP
load, where there is excessive contention on a single mutex, etc.
- I think we now have a patch that resolves this, which is being
tested by Garrett Wollman. If testing goes well and a refined
version of the patch makes it into head, this should be resolved.
- A few sites have run into performance problems for the new NFS client,
that were a result of it using a larger default I/O size (64K vs 32K
for the old one). I believe the performance problem was related to
network interface/driver issues for the larger 64K+ RPC messages.
This can be dealt with 2 ways:
- Leave it as it is now and then any site with this problem will
need to use rsize=32768,wsize=32768 mount options to avoid it.
OR
- Modify the new client so that it uses 32K by default instead of
the size specified by a server. (Most servers specify 64K or larger
these days. FreeBSD is currently at 64. Solaris is 256K and allows
1Mbyte.
So, I was going to ask the question. Maybe now is the time.
Any comments w.r.t. removing the oldNFS from the kernel for 10.0
would be appreciated, rick
> --
> Andre
> _______________________________________________
> freebsd-current@freebsd.org mailing list
> http://lists.freebsd.org/mailman/listinfo/freebsd-current
> To unsubscribe, send any mail to
> "freebsd-current-unsubscribe@freebsd.org"
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?2090268844.3936361.1363355601938.JavaMail.root>