2008-01-14 16:55:46

by Sev Binello

[permalink] [raw]
Subject: [NFS] Redhat WS4 mount default changed to tcp ?

Does anyone know the reason Redhat changed the default mount protocol to
tcp ?
Can anything be inferred from this change ?
Should we be mounting file systems tcp now ?

Thank
-Sev

--

Sev Binello
Brookhaven National Laboratory
Upton, New York
631-344-5647
[email protected]


-------------------------------------------------------------------------
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services for
just about anything Open Source.
http://ad.doubleclick.net/clk;164216239;13503038;w?http://sf.net/marketplace
_______________________________________________
NFS maillist - [email protected]
https://lists.sourceforge.net/lists/listinfo/nfs
_______________________________________________
Please note that [email protected] is being discontinued.
Please subscribe to [email protected] instead.
http://vger.kernel.org/vger-lists.html#linux-nfs



2008-01-14 17:07:17

by Trond Myklebust

[permalink] [raw]
Subject: Re: [NFS] Redhat WS4 mount default changed to tcp ?


On Mon, 2008-01-14 at 11:41 -0500, Sev Binello wrote:
> Does anyone know the reason Redhat changed the default mount protocol to
> tcp ?

It is a more robust protocol than UDP, specially when you are using
large values for rsize/wsize (by 'large' I mean 'significantly greater
than the MTU for your network').

> Can anything be inferred from this change ?
> Should we be mounting file systems tcp now ?

If you have no performance problems with UDP, then by all means keep
using it...

Cheers
Trond


-------------------------------------------------------------------------
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services for
just about anything Open Source.
http://ad.doubleclick.net/clk;164216239;13503038;w?http://sf.net/marketplace
_______________________________________________
NFS maillist - [email protected]
https://lists.sourceforge.net/lists/listinfo/nfs
_______________________________________________
Please note that [email protected] is being discontinued.
Please subscribe to [email protected] instead.
http://vger.kernel.org/vger-lists.html#linux-nfs


2008-01-14 17:29:58

by Sev Binello

[permalink] [raw]
Subject: Re: [NFS] Redhat WS4 mount default changed to tcp ?

Trond Myklebust wrote:
> On Mon, 2008-01-14 at 11:41 -0500, Sev Binello wrote:
>
>> Does anyone know the reason Redhat changed the default mount protocol to
>> tcp ?
>>
>
> It is a more robust protocol than UDP, specially when you are using
> large values for rsize/wsize (by 'large' I mean 'significantly greater
> than the MTU for your network').
>
>
Wasn't that always the case ? Was wondering Why now ?
>> Can anything be inferred from this change ?
>> Should we be mounting file systems tcp now ?
>>
>
> If you have no performance problems with UDP, then by all means keep
> using it...
>
> Cheers
> Trond
>
>


--

Sev Binello
Brookhaven National Laboratory
Upton, New York
631-344-5647
[email protected]


-------------------------------------------------------------------------
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services for
just about anything Open Source.
http://ad.doubleclick.net/clk;164216239;13503038;w?http://sf.net/marketplace
_______________________________________________
NFS maillist - [email protected]
https://lists.sourceforge.net/lists/listinfo/nfs
_______________________________________________
Please note that [email protected] is being discontinued.
Please subscribe to [email protected] instead.
http://vger.kernel.org/vger-lists.html#linux-nfs


2008-01-14 17:39:42

by Trond Myklebust

[permalink] [raw]
Subject: Re: [NFS] Redhat WS4 mount default changed to tcp ?


On Mon, 2008-01-14 at 12:22 -0500, Sev Binello wrote:
> Trond Myklebust wrote:
> > On Mon, 2008-01-14 at 11:41 -0500, Sev Binello wrote:
> >
> >> Does anyone know the reason Redhat changed the default mount protocol to
> >> tcp ?
> >>
> >
> > It is a more robust protocol than UDP, specially when you are using
> > large values for rsize/wsize (by 'large' I mean 'significantly greater
> > than the MTU for your network').
> >
> >
> Wasn't that always the case ? Was wondering Why now ?

The reason why it wasn't made a default earlier is that TCP server
support was not too good throughout most of the 2.4.x series of kernels.
It has been improved in more recent kernels to the point where everyone
feels comfortable with having it as a default.

Trond


-------------------------------------------------------------------------
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services for
just about anything Open Source.
http://ad.doubleclick.net/clk;164216239;13503038;w?http://sf.net/marketplace
_______________________________________________
NFS maillist - [email protected]
https://lists.sourceforge.net/lists/listinfo/nfs
_______________________________________________
Please note that [email protected] is being discontinued.
Please subscribe to [email protected] instead.
http://vger.kernel.org/vger-lists.html#linux-nfs