From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 25269 invoked by alias); 9 Apr 2008 20:56:07 -0000 Received: (qmail 25254 invoked by uid 9657); 9 Apr 2008 20:56:06 -0000 Date: Wed, 09 Apr 2008 20:56:00 -0000 Message-ID: <20080409205606.25252.qmail@sourceware.org> From: wysochanski@sourceware.org To: lvm-devel@redhat.com, lvm2-cvs@sourceware.org Subject: LVM2 ./WHATS_NEW test/t-vgsplit-operation.sh t ... Mailing-List: contact lvm2-cvs-help@sourceware.org; run by ezmlm Precedence: bulk List-Id: List-Subscribe: List-Post: List-Help: , Sender: lvm2-cvs-owner@sourceware.org X-SW-Source: 2008-04/txt/msg00030.txt.bz2 CVSROOT: /cvs/lvm2 Module name: LVM2 Changes by: wysochanski@sourceware.org 2008-04-09 20:56:06 Modified files: . : WHATS_NEW test : t-vgsplit-operation.sh tools : vgsplit.c Log message: Fix vgsplit to only move hidden 'snapshotN' LVs when necessary. This bug has been around for a long time as far as I can tell. Without this fix, a vgsplit would unconditionally move the 'hidden/internal' snapshot LVs, and result in corrupted metadata in the following case: vg1: contains lv1, lv1snap, both on pvset1 vg1: contains lv2, on pvset2 "vgsplit vg1 vg2 pvset2" would result in "snapshot0" hidden LV being moved to vg2, and the origin and cow being left in vg1. The tools detect the corruption in vg2, but not in vg1. Patches: http://sourceware.org/cgi-bin/cvsweb.cgi/LVM2/WHATS_NEW.diff?cvsroot=lvm2&r1=1.840&r2=1.841 http://sourceware.org/cgi-bin/cvsweb.cgi/LVM2/test/t-vgsplit-operation.sh.diff?cvsroot=lvm2&r1=1.15&r2=1.16 http://sourceware.org/cgi-bin/cvsweb.cgi/LVM2/tools/vgsplit.c.diff?cvsroot=lvm2&r1=1.57&r2=1.58 --- LVM2/WHATS_NEW 2008/04/09 14:47:34 1.840 +++ LVM2/WHATS_NEW 2008/04/09 20:56:05 1.841 @@ -1,5 +1,6 @@ Version 2.02.34 - =================================== + Fix vgsplit to only move hidden 'snapshotN' LVs when necessary. Update vgsplit tests for lvnames on the cmdline. Update vgsplit man page to reflect lvnames on the cmdline. Update vgsplit to take "-n LogicalVolumeName" on the cmdline. --- LVM2/test/t-vgsplit-operation.sh 2008/04/09 14:47:34 1.15 +++ LVM2/test/t-vgsplit-operation.sh 2008/04/09 20:56:06 1.16 @@ -149,6 +149,35 @@ lvremove -f $vg2/$lv1 && vgremove -f $vg2' +test_expect_success \ + "vgsplit correctly splits linear LV but not snap+origin LV into $i VG ($j args)" \ + 'vgcreate $vg1 $d1 $d2 && + if [ $i == existing ]; then + vgcreate $vg2 $d3 + fi && + lvcreate -l 64 -i 2 -n $lv1 $vg1 && + lvcreate -l 4 -i 2 -s -n $lv2 $vg1/$lv1 && + vgextend $vg1 $d4 && + lvcreate -l 64 -n $lv3 $vg1 $d4 && + vgchange -an $vg1 && + if [ $j == PV ]; then + vgsplit $vg1 $vg2 $d4 + else + vgsplit -n $lv3 $vg1 $vg2 + fi && + if [ $i == existing ]; then + vg_validate_pvlv_counts_ $vg2 2 1 0 + vg_validate_pvlv_counts_ $vg1 2 1 1 + else + vg_validate_pvlv_counts_ $vg2 1 1 0 + vg_validate_pvlv_counts_ $vg1 2 1 1 + fi && + lvremove -f $vg1/$lv2 && + lvremove -f $vg1/$lv1 && + lvremove -f $vg2/$lv3 && + vgremove -f $vg1 && + vgremove -f $vg2' + done done --- LVM2/tools/vgsplit.c 2008/04/09 13:47:13 1.57 +++ LVM2/tools/vgsplit.c 2008/04/09 20:56:06 1.58 @@ -164,6 +164,9 @@ return 1; } +/* + * Move the hidden / internal "snapshotN" LVs.from 'vg_from' to 'vg_to'. + */ static int _move_snapshots(struct volume_group *vg_from, struct volume_group *vg_to) { @@ -191,14 +194,21 @@ " two Volume Groups", seg->cow->name); return 0; } - } - /* Move this snapshot */ - list_del(lvh); - list_add(&vg_to->lvs, lvh); + /* + * At this point, the cow and origin should already be + * in vg_to. + */ + if (_lv_is_in_vg(vg_to, seg->cow) && + _lv_is_in_vg(vg_to, seg->origin)) { + list_del(lvh); + list_add(&vg_to->lvs, lvh); + + vg_from->snapshot_count--; + vg_to->snapshot_count++; + } + } - vg_from->snapshot_count--; - vg_to->snapshot_count++; } return 1;