fam
[Top] [All Lists]

[fam] Re: [prepatch] Directory Notification

To: Michael Gerdts <gerdts@xxxxxxxxxxxx>
Subject: [fam] Re: [prepatch] Directory Notification
From: Trond Myklebust <trond.myklebust@xxxxxxxxxx>
Date: 22 May 2000 22:22:11 +0200
Cc: "Theodore Y. Ts'o" <tytso@xxxxxxx>, willy@xxxxxxxxxxxxxxxxxx, Erez Zadok <ezk@xxxxxxxxxxxxxxx>, linux-kernel@xxxxxxxxxxxxxxxx, linux-fsdevel@xxxxxxxxxxxxxxxx, tridge@xxxxxxxxxxxxx, fam@xxxxxxxxxxx
In-reply-to: Michael Gerdts's message of "Mon, 22 May 2000 13:26:32 -0500"
References: <20000521121830.X28590@xxxxxxxxxxxxxxxxxx> <200005221225.IAA05370@xxxxxxxxxxxxxxxxx> <20000522132631.A24380@xxxxxxxxxxxx>
Sender: owner-fam@xxxxxxxxxxx
>>>>> " " == Michael Gerdts <gerdts@xxxxxxxxxxxx> writes:

     > My first thought was along these lines as well.  Sun faced a
     > similar situation with POSIX ACL's over NFS.  The solution was
     > to make it so that their nfsd added an nfs_acl service, which
     > made commands like getfacl() and setfacl() work properly over
     > NFS.

     > It seems to me that directory notifications could be handled in
     > a similar manner without a whole lot of difficulty.
     > Applications (like always) would just need to be aware that not
     > all file systems will support this feature.  Additionall, the
     > NFS client code would need to be aware that the server may not
     > support directory notifications.

This would require some sort of stateful protocol. Anything like this
would need very careful planning and design of a protocol in order not
to impose too much of a burden on the server.

It was easier with ACLs because they don't require anything beyond the
usual stateless model.

Cheers,
  Trond

--
Source code, list archive, and docs: http://oss.sgi.com/projects/fam/
To unsubscribe: echo unsubscribe fam | mail majordomo@xxxxxxxxxxx

<Prev in Thread] Current Thread [Next in Thread>