From: "Vijay Chauhan" Subject: Re: NFS mount error Date: Fri, 10 Mar 2006 16:53:17 +0530 Message-ID: References: <440F06BB.1090802@citi.umich.edu> Mime-Version: 1.0 Content-Type: multipart/alternative; boundary="----=_Part_15283_19493057.1141989797054" Cc: nfs@lists.sourceforge.net Return-path: Received: from sc8-sf-mx2-b.sourceforge.net ([10.3.1.92] helo=mail.sourceforge.net) by sc8-sf-list2.sourceforge.net with esmtp (Exim 4.30) id 1FHfiI-0003wo-O6 for nfs@lists.sourceforge.net; Fri, 10 Mar 2006 03:23:18 -0800 Received: from xproxy.gmail.com ([66.249.82.192]) by mail.sourceforge.net with esmtp (Exim 4.44) id 1FHfiI-0004oZ-Dt for nfs@lists.sourceforge.net; Fri, 10 Mar 2006 03:23:18 -0800 Received: by xproxy.gmail.com with SMTP id t5so483203wxc for ; Fri, 10 Mar 2006 03:23:17 -0800 (PST) To: cel@citi.umich.edu In-Reply-To: <440F06BB.1090802@citi.umich.edu> Sender: nfs-admin@lists.sourceforge.net Errors-To: nfs-admin@lists.sourceforge.net List-Unsubscribe: , List-Id: Discussion of NFS under Linux development, interoperability, and testing. List-Post: List-Help: List-Subscribe: , List-Archive: ------=_Part_15283_19493057.1141989797054 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline Thanks for ur explained reply. - Vijay On 3/8/06, Chuck Lever wrote: > > Vijay Chauhan wrote: > > hi, > > im doing mount/unmount in an infinte loop > > while [ "a" =3D=3D "a" ]; > > do > > mount -t nfs -o tcp server:/filepath /mount-point > > unmount -at nfs > > done > > > > after some successful mount/unmount its giving error message like > > > > RPC: can't bind to reserved port. > > > > what does it means? > > > > TIA, > > Vijay > > there are a limited number of reserved ports on every host. each mount > operation takes up to 5 ports (5 sockets for contacting the server's > portmapper, mount daemon, and so on). when each mount is finished, it > leaves 4 sockets in TIME_WAIT and one, to the NFS server, is ESTABLISHED. > > after about 120 seconds the TIME_WAIT sockets will go away, freeing > those ports. but if you mount in a tight loop, you are not giving these > sockets long enough to time out. so all the reserved ports are used up > with TIME_WAIT sockets. you can see this with the "netstat" command. > > if you wait a few minutes, you will find that the ports free up and you > can do more mounts. or you can add a "sleep" in your loop. > > there has been some mitigation in later kernels. > > 1. port numbers can be reused immediately in some circumstances. > > 2. steve d has reduced the number of sockets used for a mount request. > > 3. NFSv4 can usually skip the portmapper query. > > 4. the RPC client will attempt to connect to the default port first, > and use the portmapper only if the requested service isn't available on > the default port. > > but a tight loop like yours will still probably consume all the reserved > ports available on a client. > > > ------=_Part_15283_19493057.1141989797054 Content-Type: text/html; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline

Thanks for ur explained reply.

- Vijay

On 3/8/06, C= huck Lever <cel@citi.umich.edu= > wrote:

Vijay Chauhan wrote:
> hi,=
> im doing mount/unmount in an infinte loop
> while [ "a&= quot; =3D=3D "a" ];
> do
> mount -t nfs -o tcp server:/filepath /mount-point
&g= t; unmount -at nfs
> done
>
> after some successful mount= /unmount its giving error message like
>
> RPC: can't bind to r= eserved port.
>
> what does it means?
>
> TIA,
> Vijay
=
there are a limited number of reserved ports on every host.  = each mount
operation takes up to 5 ports (5 sockets for contacting the s= erver's
portmapper, mount daemon, and so on).  when each mount is fin= ished, it
leaves 4 sockets in TIME_WAIT and one, to the NFS server, is E= STABLISHED.

after about 120 seconds the TIME_WAIT sockets will go aw= ay, freeing
those ports.  but if you mount in a tight loop, you are not g= iving these
sockets long enough to time out.  so all the reser= ved ports are used up
with TIME_WAIT sockets.  you can see thi= s with the "netstat" command.

if you wait a few minutes, you will find that the ports free up and= you
can do more mounts.  or you can add a "sleep" i= n your loop.

there has been some mitigation in later kernels.
1.  port numbers can be reused immediately in some circumstances= .

2.  steve d has reduced the number of sockets used for a = mount request.

3.  NFSv4 can usually skip the portmapper q= uery.

4.  the RPC client will attempt to connect to the de= fault port first,
and use the portmapper only if the requested service i= sn't available on
the default port.

but a tight loop like yours will still probabl= y consume all the reserved
ports available on a client.

------=_Part_15283_19493057.1141989797054-- ------------------------------------------------------- This SF.Net email is sponsored by xPML, a groundbreaking scripting language that extends applications into web and mobile media. Attend the live webcast and join the prime developer group breaking into this new coding territory! http://sel.as-us.falkag.net/sel?cmd=lnk&kid=110944&bid=241720&dat=121642 _______________________________________________ NFS maillist - NFS@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/nfs