rlaager

--- zfs-linux-0.6.0.44/debian/tree/usr/share/initramfs-tools/scripts/zfs	2011-01-12 19:30:38.000000000 -0600
+++ zfs-linux-0.6.0.44/debian/tree/usr/share/initramfs-tools/scripts/zfs	2012-01-15 18:34:53.985778000 -0600
@@ -64,15 +64,22 @@
 	# environment and doing a full import in the regular system instead.
 
 	[ "$quiet" != "y" ] && log_begin_msg "Importing ZFS root pool $ZFS_RPOOL"
-	if [ -f /etc/zfs/zpool.cache ]
-	then
-		ZFS_STDERR=$(zpool list "$ZFS_RPOOL" 1>/dev/null 2>&1 \
-		  || zpool import -f -N "$ZFS_RPOOL" 2>&1)

rlaager

commit 837f51c82ca25a3a570114c2f308398e04caee11
Author: Richard Laager <[email protected]>
Date:   Sat Feb 25 20:22:44 2012 -0600

    raw-posix: Discard using fallocate() or BLKDISCARD
    
    The fallocate(fd, FALLOC_FL_PUNCH_HOLE) call is used for regular files,
    while ioctl(fd, BLKDISCARD) is used for devices.  Thus, discard is no
    longer XFS-only.

rlaager

#include <sys/types.h>
#include <sys/stat.h>
#include <fcntl.h>
#include <unistd.h>
#include <stdio.h>
#include <stdlib.h>
#include <linux/falloc.h>

int main()

rlaager

ashift=9 vdev=raidz1 disks=3
tank/foo   102M  1.84G   101M  -

ashift=9 vdev=raidz1 disks=4
tank/foo   103M  2.80G   103M  -

ashift=9 vdev=raidz1 disks=5
tank/foo   102M  3.76G   101M  -

ashift=9 vdev=raidz1 disks=6

rlaager

ZFS on Linux Improvements

1. `zpool status` can output disk names which are under /dev/disk.
2. `zpool status` outputs the whole disk device for wholedisk pools,
   but GRUB needs the partition device.

Before this can be accepted upstream, I am to merge it into the
existing /dev scanning code.

Index: grub/util/getroot.c

rlaager

diff -r 10ae2f408c94 dice/dice.c
--- a/dice/dice.c	Wed Jan 18 22:20:42 2012 -0600
+++ b/dice/dice.c	Sun Jun 10 16:17:12 2012 -0500
@@ -30,6 +30,10 @@
 #include <debug.h>
 #include <plugin.h>
 
+#define PLUGIN_ID		"core-plugin_pack-dice"
+#define PREFS_PREFIX		"/plugins/core/" PLUGIN_ID
+#define PREFS_REMOTE_CMD	PREFS_PREFIX "/remote_cmd"

rlaager

For the c2t5000C5002623694Bd0 disk, we received a SMART predictive failure
(fault.io.disk.predictive-failure) warning as well as
fault.io.scsi.cmd.disk.dev.rqs.derr from fmadm.

We then set out to replace the disk.

Here is the zpool status before the replacement:

# zpool status data
  pool: data

rlaager

$ xdpyinfo
name of display:    :0.0
version number:    11.0
vendor string:    The X.Org Foundation
vendor release number:    11103000
X.Org version: 1.11.3
maximum request size:  16777212 bytes
motion buffer size:  256
bitmap unit, bit order, padding:    32, LSBFirst, 32
image byte order:    LSBFirst

rlaager

hdparm -I /dev/sda
# Note that it is "not locked" and security is "not enabled".

# Set a password:
hdparm --user-master u --security-set-pass password /dev/sda

# From this point, until after the security erase, if something goes wrong,
# you must remove the password before rebooting.

hdparm -I /dev/sda'

rlaager

# Add a default partition map on the new drive.
fdisk -B c7t1d0p0

# Copy the disk label from the first disk to the second:
prtvtoc /dev/rdsk/c7t0d0s0 | fmthard -s - /dev/rdsk/c7t1d0s0

# Attach the second disk to the root pool:
zpool attach rpool c7t0d0s0 c7t1d0s0
# You will get an error about overlapping. If you get another error, stop. Otherwise:
zpool attach -f rpool c7t0d0s0 c7t1d0s0