Skip site navigation (1)Skip section navigation (2)
Date:      Mon, 14 Dec 1998 07:36:05 +0100
From:      Bernd Walter <ticso@cicely.de>
To:        Kevin Day <toasty@home.dragondata.com>, hackers@FreeBSD.ORG
Subject:   Re: NFS thoughts
Message-ID:  <19981214073605.20458@cicely.de>
In-Reply-To: <199812132342.RAA02261@home.dragondata.com>; from Kevin Day on Sun, Dec 13, 1998 at 05:42:45PM -0600
References:  <199812132342.RAA02261@home.dragondata.com>

next in thread | previous in thread | raw e-mail | index | archive | help
On Sun, Dec 13, 1998 at 05:42:45PM -0600, Kevin Day wrote:
> 
> I've noticed a few things about NFS. I don't know enough about this to be
> useful, so this is a lot of guesswork.
> 
> I get my nfs client complaining about the server not responding a lot. I
> don't see how this is possible, because the server is connected over 100MB
> ethernet, on a very not busy segment, and the server isn't too busy.
> 
> This began to make me wonder.... The dynamic retransmit algorithm doesn't
> look like it was meant for very high speed links like this.
> 
> If the last few calls went very very quickly(which they appear to be), just
> a few collisions alone could knock this above the retransmit limit.
> 
> I also seem to see this happening a lot right at 3am, when a big cron job
> goes off on the server, making the replies come in later. Mounting with -d
> seems to help this, but I'm going to experiment with changing the algorithm
> to back off in much bigger steps.
> 
> Rpc Info:
>  TimedOut   Invalid X Replies   Retries  Requests
>         4         0      4749     10270   5587819
> 
> Usually, after each TimedOut that's occured, 5-10 processes seem to randomly
> exit on SIGSEGV. This is another issue that seems seperate, but if I can
> prevent timeouts from occuring anyway, this would go away too.
> 
> Has anyone ventured down this path already?
I saw the same on my private hosts.
Everythings the same to your case instead that I have a 100MBit FreeBSD Router
between them. All lines are running Full-Duplex Point-to-Point.
In my case I have a syslogentry telling me about a server down under some load
and it took minutes till it says that the server is up again.
It happend when using NFS3/TCP at this moment I'm using NFS2/UDP and it won't
hang.
Another issue is that when using NFS with multihomed hosts the client ask on
one IP address of the server and the server replies using another of his IPs,
so the client is discarding the answers and still waiting.
> 
> 
> Kevin
> 
> To Unsubscribe: send mail to majordomo@FreeBSD.org
> with "unsubscribe freebsd-hackers" in the body of the message

-- 
  B.Walter


To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-hackers" in the body of the message



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?19981214073605.20458>