Differences

This shows you the differences between two versions of the page.

--- nexenta-gnusolaris [2008/02/04 15:37]
pegasus created
+++ nexenta-gnusolaris [2012/10/21 19:40] (current)
a
@@ Line 1: / Line 1: @@
-First things first - Nexenta is NOT debain with opensolaris kernel, it's opensolaris with a few debian tools, like apt. It's anoying for a linux admin, because ps ax is ps -ef, mount -t is mount -F, ifconfig barfs without parametrs etc.
+====== Nexenta ======
-But at least bash is the default shell :)
+<note important>First things first - **Nexenta** is NOT debain with opensolaris kernel, it's __opensolaris with a few debian tools__, like apt. It's anoying for a linux admin, because ps ax is ps -ef, mount -t is mount -F, ifconfig barfs without parametrs etc.
+But at least bash is the default shell :)</note>
 Other changes, that requires some adjusting to - SMF. Yes, init.d and inittab are history. See [[http://docs.sun.com/app/docs/doc/817-1985/6mhm8o5rl?a=view|this chapter]] of [[http://docs.sun.com/app/docs/coll/47.16|documentation]].
@@ Line 10: / Line 12: @@
 So the answer is to install manually on one disk, where you create a small (eg 20gb) root slice, 1gb swap slice and the rest for /export/home or whatever. Let the install finish, reboot, format the other drive that will become a mirror to the one you installed on (warning, solaris slices! :) and then use zpool attach <poolname> <existing device> <new device> to create a mirror. What remains is just installgrub on the new device and voila, you have a mirrored bootable zfs root gnu/opensolaris system.
-Few tricks: if format complains about not being able to relabel a disk that is in use, simply export NOINUSE_CHECK=1. Also when creating a zpool from slices, it complains that s0 overlaps with s2 (which should be that way, if I understand slices corectly), but we can simply use zpool create -f. After you created your zpool and zfs volumes to your desire, be sure to check zfs get all <name> and check what all can you set on them. I found compression=on and atime=off very userful :)
+Few tricks: if format complains about not being able to relabel a disk that is in use, simply export NOINUSE_CHECK=1. Also when creating a zpool from slices, it complains that s0 overlaps with s2 (which should be that way, if I understand slices corectly), but we can simply use zpool create -f. After you created your zpool and zfs volumes to your desire, be sure to check zfs get all <name> and check what all can you set on them. I found compression=on and atime=off very userful. Also set snapdir=visible if you plan to offer backup history via shapshots. It's amazingly simple :)
+**Useful links**
+   * [[:zfs|ZFS Filesystem on TnT]]
+   * [[http://www.cupfighter.net/index.php/2012/10/default-nexenta-zfs-settings-you-want-to-change/|Default Nexenta / ZFS settings you want to change]]