From: Brent (brent@eng.sun.com)
Date: 03/18/99-02:29:25 PM Z
Message-ID: <36F16225.658AEEDF@eng.sun.com> Date: Thu, 18 Mar 1999 12:29:25 -0800 From: Brent <brent@eng.sun.com> Subject: Re: fsid.major/minor "Noveck, Dave" wrote: > > df reports a file system space avail value that is very > > relevant - the amount of free space in the quota tree, if > > the user mounts the quota tree itself (as opposed to the root > > of the entire file system). > > Cute. > > It seems to me that in NFS-v4 as currently defined, you would > not be able to do the same thing. In v2-v3 you know what the > client is mounting; you get that from the mount protocol. In > v4, the server just does lookups relative to the server's root > (mod all the stuff about dummy fs's for non-exported stuff). > But is there a way for the server to know what the client is > mounting? No, the concept of a client mount is not important to the protocol. The client can lookup any filehandle and decide that it'll be the root of a mounted filesystem - unbeknownst to the server. The MOUNT protocol does have a fuzzy concept of client mounts (which we record in the ugly /etc/mtab file) but I'll be happy if NFSv4 puts an end to this. I don't really see a problem here though - if the client wants to know how much space is available on the filesystem identified by fsid, then it uses free_space attribute (UNIX f_bavail). There's also an avail_space attribute that identifies the space available to "this" user (UNIX f_favail) - which may be smaller. This is independent of the quota information that's available with the quota_* attributes. What each of these attributes applies to depends on the current filehandle, and the definition of the attribute. What the client does with it is of no consequence. Brent
This archive was generated by hypermail 2.1.2 : 03/04/05-01:46:52 AM Z CST