Attachment 446065 Details for Bug 712539

[patch] Patch to move dmeventd start after local filesystem mounting

rootonlvm_dmeventd_errmsgs.diff (text/plain), 3.05 KB, created by Pieter Hollants on 2011-08-16 22:27:58 UTC

(hide)

Description:

Filename:

MIME Type:

Creator: Pieter Hollants

Created: 2011-08-16 22:27:58 UTC

Size: 3.05 KB

patch

obsolete

># If root is on a LVM LV, we get a number of error messages on boot
># about "failure to start dmeventd", "event registration failed" etc.
># This is due to dmeventd now being part of device-mapper and 
># automatically started to monitor eg. RAID-1 style devices with the 
># "mirror" plugin.
>#
># However, dmeventd is a.) not part of the initrd and b.) requires /var
># and possibly other resources. Generally, it does not make sense to
># enable monitoring until all local filesystems have available. vgchange
># got a --sysinit parameter added because of this.
>#
># This patch:
># - adds "--sysinit" to the initrd boot-lvm2.sh script responsible of
>#   making LVM VGs available (not even a root filesystem available at 
>#   this stage)
># - adds "--sysinit" to the full-blown boot.lvm start script (if eg. 
>#   /var is on LVM, vgchange will yield errors here, too)
># - adds a new /etc/init.d/boot.lvm_monitor script that enables 
>#   monitoring just after local filesystems have become available
>#
># This is the same behaviour that eg. Arch Linux and Gentoo have 
># adopted (Credits to them for the idea behind the fix).
>#
># Patch by Pieter Hollants <pieter@hollants.com>
>--- /lib/mkinitrd/scripts//boot-lvm2.sh	2011-02-19 06:17:45.000000000 +0100
>+++ /lib/mkinitrd/scripts//boot-lvm2.sh	2011-08-16 23:36:10.033172599 +0200
>@@ -72,7 +72,7 @@
> 
> # initialize remebered and parameterized devices
> for vgr in $vg_root $vg_resume $vg_roots; do
>-	vgchange -a y $vgr
>+	vgchange -a y $vgr --sysinit
> done
> 
> unset lvm2_get_vg
>--- /etc/init.d/boot.lvm	2011-02-19 06:17:45.000000000 +0100
>+++ /etc/init.d/boot.lvm	2011-08-16 23:59:59.194126823 +0200
>@@ -40,8 +40,11 @@
> 	    fi
> 	    echo "Scanning for LVM volume groups..."
> 	    /sbin/vgscan --mknodes
>+
>+            # --sysinit because not all filesystems might be available
>+            # rw yet (eg. /var on LVM)
> 	    echo "Activating LVM volume groups..."
>-	    /sbin/vgchange -a y $LVM_VGS_ACTIVATED_ON_BOOT
>+	    /sbin/vgchange -a y --sysinit $LVM_VGS_ACTIVATED_ON_BOOT
> 
> 	    rc_status -v -r
> 	fi
>--- /etc/init.d/boot.lvm_monitor	1970-01-01 01:00:00.000000000 +0100
>+++ /etc/init.d/boot.lvm_monitor	2011-08-16 23:54:09.424803351 +0200
>@@ -0,0 +1,46 @@
>+#! /bin/sh
>+#
>+# Copyright (c) 2001 SuSE GmbH Nuernberg, Germany.  All rights reserved.
>+#
>+# /etc/init.d/boot.lvm_monitor
>+#
>+### BEGIN INIT INFO
>+# Provides:          boot.lvm_monitor
>+# Required-Start:    $local_fs
>+# Should-Start:      $null
>+# Required-Stop:     $null
>+# Should-Stop:       $null
>+# Default-Start:     B
>+# Default-Stop:
>+# Description:       start monitoring of LVM VGs now filesystems are mounted rw
>+### END INIT INFO
>+
>+. /etc/rc.status
>+. /etc/sysconfig/lvm
>+
>+rc_reset
>+
>+case "$1" in
>+  start)
>+	if test -d /etc/lvm -a -x /sbin/vgscan -a -x /sbin/vgchange ; then
>+	    echo "Enabling monitoring on LVM volume groups..."
>+	    /sbin/vgchange --monitor y $LVM_VGS_ACTIVATED_ON_BOOT
>+	    rc_status -v -r
>+	fi
>+	;;
>+    stop)
>+	rc_status -v
>+	;;
>+    status)
>+	rc_status -v
>+	;;
>+    reload)
>+	rc_status -v
>+	;;
>+    *)
>+	echo "Usage: $0 {start|stop|status|reload}"
>+	exit 1
>+	;;
>+esac
>+
>+rc_exit

Actions: View | Diff

Attachments on bug 712539: 446065