Re: Mount point crossing & space_avail

New Message Reply About this list Date view Thread view Subject view Author view Attachment view

From: Peter Åstrand (peter@cendio.se)
Date: 03/20/02-01:36:26 PM Z


Date: Wed, 20 Mar 2002 20:36:26 +0100 (CET)
From: Peter Åstrand <peter@cendio.se>
Subject: Re: Mount point crossing & space_avail 
Message-ID: <Pine.LNX.4.44.0203202024560.15634-100000@sofie.lambo.student.liu.se>

> I would be opposed to adding a "list exported file systems" op.  Current
> file systems tend to be fairly static, but I could imagine a server that
> builds its tree dynamically.  Plan 9, for example, allows you to build
> arbitrarily complex name spaces.  

A Plan 9 server can return NFS4ERR_NOTSUPP for a "list exported file 
systems" op, if it is unable to return any information that makes sense. 

We already have lots of ops that can return NOTSUPP, like OPENATTR. 

(Btw, why can operations like OPEN and RENAME return NOTSUPP? Aren't these 
mandatory? If NOTSUPP are allowed for OPEN, shouldn't it be allowed for 
LOCK etc as well?)

> Afs solves this problem by providing a separate command that you can run
> on any directory to find out how much space is available in that file
> system. If you have this, which can be supported by the current v4
> protocol, I don't think you need a way to query the server for a list of
> file systems.

This command then walks the entire directory tree, right? Yes, that would
work. But it does not scale very well. The response time is proportional
to the number of directories in the tree. A slightly to ugly solution to 
be usable by an ordinary "df" command, IMHO. 


-- 
Peter Åstrand                Telephone: +46-13-21 46 00
Cendio Systems               E-mail: peter@cendio.se
Teknikringen 3
583 30 Linköping
Sweden


New Message Reply About this list Date view Thread view Subject view Author view Attachment view

This archive was generated by hypermail 2.1.2 : 03/04/05-01:49:37 AM Z CST