Fix bitmap migration
cover Fix bitmap migration
Hi all!
We've faced a bug in rhev-2.12.0-33.el7-based Qemu.
In upstream, bug introduced in 4.0 by 74da6b943565c45 "block/dirty-bitmaps: implement inconsistent bit" commit. At this commit we started to load inconsistent bitmap instead of silently ignoring them, and it now I see that it breaks migration.
The fix is very simple, so I think it's OK for 4.2.. Still, it's not a degradation, so we may postpone it to 5.0.
Vladimir Sementsov-Ogievskiy (2):
block/qcow2-bitmap: fix bitmap migration
iotests: add new test cases to bitmap migration
block/qcow2-bitmap.c | 21 ++++++++++++++++++++- tests/qemu-iotests/169 | 22 +++++++++++++++------- tests/qemu-iotests/169.out | 4 ++-- 3 files changed, 37 insertions(+), 10 deletions(-)
Signature · 1 line
2.21.0
PATCH 1/2 block/qcow2-bitmap: fix bitmap migration
Fix bitmap migration with dirty-bitmaps capability enabled and shared storage. We should ignore IN_USE bitmaps in the image on target, when migrating bitmaps through migration channel, see new comment below.
--- block/qcow2-bitmap.c | 21 ++++++++++++++++++++- 1 file changed, 20 insertions(+), 1 deletion(-)
block/qcow2-bitmap.c +20 -1
@@ -988,7 +988,26 @@ bool qcow2_load_dirty_bitmaps(BlockDriverState *bs, Error **errp) } QSIMPLEQ_FOREACH(bm, bm_list, entry) { -BdrvDirtyBitmap *bitmap =load_bitmap (bs, bm, errp); +BdrvDirtyBitmap *bitmap; + +if ((bm->flags &BME_FLAG_IN_USE ) && +bdrv_find_dirty_bitmap (bs, bm->name)) + { +/* +* We already have corresponding BdrvDirtyBitmap, and bitmap in the +* image is marked IN_USE. Firstly, this state is valid, no reason +* to consider existing BdrvDirtyBitmap to be bad. Secondly it's +* absolutely possible, when we do migration with shared storage +* with dirty-bitmaps capability enabled: if the bitmap was loaded +* from this storage before migration start, the storage will +* of-course contain IN_USE outdated version of the bitmap, and we +* should not load it on migration target, as we already have this +* bitmap, being migrated. +*/ +continue ; + } + + bitmap =load_bitmap (bs, bm, errp);if (bitmap ==NULL ) { goto fail; }
Signature · 1 line
2.21.0
PATCH 2/2 iotests: add new test cases to bitmap migration
In reply to Vladimir Sementsov-OgievskiyAdd optional pre-shutdown: shutdown/launch vm before migration. This leads to storing persistent bitmap to the storage, which breaks migration with dirty-bitmaps capability enabled and shared storage until fixed by previous commit.
--- tests/qemu-iotests/169 | 22 +++++++++++++++------- tests/qemu-iotests/169.out | 4 ++-- 2 files changed, 17 insertions(+), 9 deletions(-)
tests/qemu-iotests/169 +15 -7
@@ -134,7 +134,7 @@ class TestDirtyBitmapMigration(iotests.QMPTestCase): self.check_bitmap(self.vm_a, sha256 if persistent else False) def do_test_migration(self, persistent, migrate_bitmaps, online, - shared_storage): + shared_storage, pre_shutdown): granularity = 512 # regions = ((start, count), ...)
@@ -142,15 +142,13 @@ class TestDirtyBitmapMigration(iotests.QMPTestCase): (0xf0000, 0x10000), (0xa0201, 0x1000)) - should_migrate = migrate_bitmaps or persistent and shared_storage + should_migrate = \ + (migrate_bitmaps and (persistent or not pre_shutdown)) or \ + (persistent and shared_storage) mig_caps = [{'capability': 'events', 'state': True}] if migrate_bitmaps: mig_caps.append({'capability': 'dirty-bitmaps', 'state': True}) - result = self.vm_a.qmp('migrate-set-capabilities', - capabilities=mig_caps) - self.assert_qmp(result, 'return', {}) - self.vm_b.add_incoming(incoming_cmd if online else "defer") self.vm_b.add_drive(disk_a if shared_storage else disk_b)
@@ -166,6 +164,14 @@ class TestDirtyBitmapMigration(iotests.QMPTestCase): self.vm_a.hmp_qemu_io('drive0', 'write %d %d' % r) sha256 = self.get_bitmap_hash(self.vm_a) + if pre_shutdown: + self.vm_a.shutdown() + self.vm_a.launch() + + result = self.vm_a.qmp('migrate-set-capabilities', + capabilities=mig_caps) + self.assert_qmp(result, 'return', {}) + result = self.vm_a.qmp('migrate', uri=mig_cmd) while True: event = self.vm_a.event_wait('MIGRATION')
@@ -210,11 +216,13 @@ def inject_test_case(klass, name, method, *args, **kwargs): mc = operator.methodcaller(method, *args, **kwargs) setattr(klass, 'test_' + method + name, lambda self: mc(self)) -for cmb in list(itertools.product((True, False), repeat=4)): +for cmb in list(itertools.product((True, False), repeat=5)): name = ('_' if cmb[0] else '_not_') + 'persistent_' name += ('_' if cmb[1] else '_not_') + 'migbitmap_' name += '_online' if cmb[2] else '_offline' name += '_shared' if cmb[3] else '_nonshared' + if (cmb[4]): + name += '__pre_shutdown' inject_test_case(TestDirtyBitmapMigration, name, 'do_test_migration', *list(cmb))
tests/qemu-iotests/169.out +2 -2
@@ -1,5 +1,5 @@ -.................... +.................................... ---------------------------------------------------------------------- -Ran 20 tests +Ran 36 tests OK
Signature · 1 line
2.21.0
Quoted 11 lines · Vladimir Sementsov-Ogievskiy
Hi all!
We've faced a bug in rhev-2.12.0-33.el7-based Qemu.
In upstream, bug introduced in 4.0 by74da6b943565c45"block/dirty-bitmaps: implement inconsistent bit" commit. At this commit we started to load inconsistent bitmap instead of silently ignoring them, and it now I see that it breaks migration.The fix is very simple, so I think it's OK for 4.2.. Still, it's not a degradation, so we may postpone it to 5.0.
Quoted 8 lines
Vladimir Sementsov-Ogievskiy (2):
block/qcow2-bitmap: fix bitmap migration
iotests: add new test cases to bitmap migrationblock/qcow2-bitmap.c | 21 ++++++++++++++++++++- tests/qemu-iotests/169 | 22 +++++++++++++++------- tests/qemu-iotests/169.out | 4 ++-- 3 files changed, 37 insertions(+), 10 deletions(-)
Makes sense to me to put this into 4.2, but I don’t think it would survive Peter’s check list. :?
(https://lists.nongnu.org/archive/html/qemu-block/2019-11/msg00807.html)
Max
Quoted 21 lines · Max Reitz
Quoted 11 lines · Vladimir Sementsov-Ogievskiy
Hi all!
We've faced a bug in rhev-2.12.0-33.el7-based Qemu.
In upstream, bug introduced in 4.0 by74da6b943565c45"block/dirty-bitmaps: implement inconsistent bit" commit. At this commit we started to load inconsistent bitmap instead of silently ignoring them, and it now I see that it breaks migration.The fix is very simple, so I think it's OK for 4.2.. Still, it's not a degradation, so we may postpone it to 5.0.
Quoted 8 lines
Vladimir Sementsov-Ogievskiy (2):
block/qcow2-bitmap: fix bitmap migration
iotests: add new test cases to bitmap migrationblock/qcow2-bitmap.c | 21 ++++++++++++++++++++- tests/qemu-iotests/169 | 22 +++++++++++++++------- tests/qemu-iotests/169.out | 4 ++-- 3 files changed, 37 insertions(+), 10 deletions(-)
Quoted 4 lines
Reviewed-by Max Reitz <[email protected]>Makes sense to me to put this into 4.2, but I don’t think it would survive Peter’s check list. :?
It depends on how much the fixed case is popular..
Actually, anyone who tries to migrate bitmaps with dirty-bitmaps capability enabled and with shared storage will very possibly run into this bug.
But who is it, except for Virtuozzo, I don't know:) Still, I see that a lot of that staff merged into Rhel Qemu we based on, so at least this should go to qemu-stable and to Rhel update.
Quoted 5 lines
(https://lists.nongnu.org/archive/html/qemu-block/2019-11/msg00807.html)
Max
Signature · 2 lines
Best regards, Vladimir
Max Reitz
AppliedQuoted 11 lines · Vladimir Sementsov-Ogievskiy
Hi all!
We've faced a bug in rhev-2.12.0-33.el7-based Qemu.
In upstream, bug introduced in 4.0 by74da6b943565c45"block/dirty-bitmaps: implement inconsistent bit" commit. At this commit we started to load inconsistent bitmap instead of silently ignoring them, and it now I see that it breaks migration.The fix is very simple, so I think it's OK for 4.2.. Still, it's not a degradation, so we may postpone it to 5.0.
Quoted 8 lines
Vladimir Sementsov-Ogievskiy (2):
block/qcow2-bitmap: fix bitmap migration
iotests: add new test cases to bitmap migrationblock/qcow2-bitmap.c | 21 ++++++++++++++++++++- tests/qemu-iotests/169 | 22 +++++++++++++++------- tests/qemu-iotests/169.out | 4 ++-- 3 files changed, 37 insertions(+), 10 deletions(-)
Thanks, applied to my block branch:
https://git.xanclic.moe/XanClic/qemu/commits/branch/block
Max