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: 03/20/02-09:34:49 AM Z


Date: Wed, 20 Mar 2002 16:34:49 +0100 (CET)
From: Peter Åstrand <peter@cendio.se>
Subject: Mount point crossing & space_avail
Message-ID: <Pine.LNX.4.44.0203201517450.7744-100000@sofie.lambo.student.liu.se>

Suppose we have a server S with three local filesystems:

/dev/disk1  /
/dev/disk2  /export/  (2 MiB bytes free)
/dev/disk3  /export/pictures/  (2 GiB bytes free)

The /export tree is exported via v4. The /export directory contains one
directory called "documents" (in addition to "pictures", where disk3 is
mounted). 

Client C mounts S:/export on /mnt. On the client, it looks like this:

$ cd /mnt
$ ls
documents  pictures
$ 

User U at client C is planning to copy 500 MiB of data to /mnt/pictures
and first checks if there is enough room for it:

$ df
Filesystem            Size  Used Avail Use% Mounted on
...
S:/export             ...   ...     2M  ... /mnt

User U, which is totally unaware of the servers disk configuration, fsids 
and "All That Jazz :-)" screams and think it's time to buy more disks. U 
cannot now that "pictures" actually is on another filesystem. Or have I 
forgotten something?


Wouldn't it be nice to have some way of querying the server of exported
filesystems, e.g. add the functionality of the MOUNT protocols EXPORT
procedure into v4? Some person asked if there was similiar functionality
in v4 at Cthon, and if I remember it correctly the answer was "No, but
most people will continue to run MOUNTD even with v4". I don't think
that's a good solution.

If v4 had some way of querying the server for exported fs's, an
NFSv4-aware df program could be a bit more intelligent:

$ df
Filesystem            Size  Used Avail Use% Mounted on
...
S:/export             ...   ...     2M  ... /mnt
S:/export/pictures    ...   ...     2G  ... (not mounted)

(Maybe it's a bad idea with a df program fetching information not only by
calling statfs(2) but also by contacting server S over the network, but at
least it would be possible.)


But I suppose an EXPORTS operation in v4 will have to wait until 4.1, 
even if we agree on that it's a Good Thing?


-- 
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 : 03/04/05-01:49:36 AM Z CST