Darren J Moffat wrote:
So I setup the share manually in /etc/dfs/dfstab thus :

# vi /etc/dfs/dfstab
"/etc/dfs/dfstab" 1 lines, 97 characters
share -F nfs -o ro=isis,root=isis -d "jumpstart"
/export/zfs_0/jumpstart/s10/SXCRb35

So why did you do that rather than what it suggested you should do which is:

# zfs set sharenfs=ro=isis,root=isis zfs_0


Because I do not want to share out the whole ZFS filesystem

Strictly /export/zfs_0/jumpstart/s10/SXCRb35

So why not have /export/zfs_0/
        /export/zfs_0/jumpstart
        /export/zfs_0/jumpstart/s10
        /export/zfs_0/jumpstart/s10/SXCRb35

all as separate ZFS filesystems, they are cheap after all :-)

ZFS filesystems are designed to be plentiful as water. Use them liberally. :) It takes awhile to get used to the new administrative model, which is very different from the traditional disk-oriented UNIX mindset.

FWIW, on my workstation I have created a separate ZFS file system for every single directory which contains data which isn't NFS mounted.

[EMAIL PROTECTED] zfs list
NAME                   USED  AVAIL  REFER  MOUNTPOINT
junk                  63.0G   157G   151K  /junk
junk/builds           57.6G   157G   151K  /builds
junk/builds/elowe     57.6G   157G  57.6G  /builds/elowe
junk/iso               121M   157G   121M  /junk/iso
junk/local            4.62M   157G  4.62M  /usr/local
junk/opt              23.9M   157G  23.9M  /junk/opt
junk/twilight         5.13G   157G  4.91G  /zone/twilight
junk/[EMAIL PROTECTED]   232M      -  4.85G  -

/junk is a ZFS version of /tmp where I put junk; /builds contains workspaces and each user gets their own filesystem for their workspaces so each is a separate ZFS filesystem. That structure is analogous to what you want. I created /junk/iso on a whim so I could export downloaded ISO images to another machine with a working CD burner. My zone is its own ZFS filesystem I installed then immediately took a snapshot. This lets me create a new zone simply by cloning the initial installation.

The side effects I like are that I can see at a glance (without du or df) where all of my disk space is going, control NFS exports via zfs commands (no more poking at /etc/dfs for me, thanks) and moreover when I decide I don't want those ISO images anymore (which I just did) I can destroy the filesystem instead of doing a rm -rf:

# zfs destroy junk/iso

The ZFS model of administration is just so much better...

- Eric
_______________________________________________
opensolaris-discuss mailing list
opensolaris-discuss@opensolaris.org

Reply via email to