This patch extends the blockdev-backup QMP command to allow users to specify
how to behave when IO errors occur during copy-before-write operations.
Previously, the behavior was fixed and could not be controlled by the user.
The new 'on-cbw-error' option can be set to one of two values:
- 'break-guest-write': Forwards the IO error to the guest and triggers
the on-source-error policy. This preserves snapshot integrity at the
expense of guest IO operations.
- 'break-snapshot': Allows the guest OS to continue running normally,
but invalidates the snapshot and aborts related jobs. This prioritizes
guest operation over backup consistency.
This enhancement provides more flexibility for backup operations in different
environments where requirements for guest availability versus backup
consistency may vary.
The default behavior remains unchanged to maintain backward compatibility.
Signed-off-by: Raman Dzehtsiar <Raman.Dzehtsiar@gmail.com>
Message-ID: <
20250414090025.828660-1-Raman.Dzehtsiar@gmail.com>
Acked-by: Markus Armbruster <armbru@redhat.com>
[vsementsov: fix long lines]
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
Tested-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
BackupPerf *perf,
BlockdevOnError on_source_error,
BlockdevOnError on_target_error,
+ OnCbwError on_cbw_error,
int creation_flags,
BlockCompletionFunc *cb, void *opaque,
JobTxn *txn, Error **errp)
}
cbw = bdrv_cbw_append(bs, target, filter_node_name, discard_source,
- perf->min_cluster_size, &bcs, errp);
+ perf->min_cluster_size, &bcs, on_cbw_error, errp);
if (!cbw) {
goto error;
}
bool discard_source,
uint64_t min_cluster_size,
BlockCopyState **bcs,
+ OnCbwError on_cbw_error,
Error **errp)
{
BDRVCopyBeforeWriteState *state;
}
qdict_put_str(opts, "file", bdrv_get_node_name(source));
qdict_put_str(opts, "target", bdrv_get_node_name(target));
+ qdict_put_str(opts, "on-cbw-error", OnCbwError_str(on_cbw_error));
if (min_cluster_size > INT64_MAX) {
error_setg(errp, "min-cluster-size too large: %" PRIu64 " > %" PRIi64,
bool discard_source,
uint64_t min_cluster_size,
BlockCopyState **bcs,
+ OnCbwError on_cbw_error,
Error **errp);
void bdrv_cbw_drop(BlockDriverState *bs);
0, MIRROR_SYNC_MODE_NONE, NULL, 0, false, false,
NULL, &perf,
BLOCKDEV_ON_ERROR_REPORT,
- BLOCKDEV_ON_ERROR_REPORT, JOB_INTERNAL,
+ BLOCKDEV_ON_ERROR_REPORT,
+ ON_CBW_ERROR_BREAK_GUEST_WRITE,
+ JOB_INTERNAL,
backup_job_completed, bs, NULL, &local_err);
if (local_err) {
error_propagate(errp, local_err);
BdrvDirtyBitmap *bmap = NULL;
BackupPerf perf = { .max_workers = 64 };
int job_flags = JOB_DEFAULT;
+ OnCbwError on_cbw_error = ON_CBW_ERROR_BREAK_GUEST_WRITE;
if (!backup->has_speed) {
backup->speed = 0;
job_flags |= JOB_MANUAL_DISMISS;
}
+ if (backup->has_on_cbw_error) {
+ on_cbw_error = backup->on_cbw_error;
+ }
+
job = backup_job_create(backup->job_id, bs, target_bs, backup->speed,
backup->sync, bmap, backup->bitmap_mode,
backup->compress, backup->discard_source,
&perf,
backup->on_source_error,
backup->on_target_error,
+ on_cbw_error,
job_flags, NULL, NULL, txn, errp);
return job;
}
* all ".has_*" fields are ignored.
* @on_source_error: The action to take upon error reading from the source.
* @on_target_error: The action to take upon error writing to the target.
+ * @on_cbw_error: The action to take upon error in copy-before-write operations.
* @creation_flags: Flags that control the behavior of the Job lifetime.
* See @BlockJobCreateFlags
* @cb: Completion function for the job.
BackupPerf *perf,
BlockdevOnError on_source_error,
BlockdevOnError on_target_error,
+ OnCbwError on_cbw_error,
int creation_flags,
BlockCompletionFunc *cb, void *opaque,
JobTxn *txn, Error **errp);
# default 'report' (no limitations, since this applies to a
# different block device than @device).
#
+# @on-cbw-error: policy defining behavior on I/O errors in
+# copy-before-write jobs; defaults to break-guest-write. (Since 10.1)
+#
# @auto-finalize: When false, this job will wait in a PENDING state
# after it has finished its work, waiting for @block-job-finalize
# before making any block graph changes. When true, this job will
'*compress': 'bool',
'*on-source-error': 'BlockdevOnError',
'*on-target-error': 'BlockdevOnError',
+ '*on-cbw-error': 'OnCbwError',
'*auto-finalize': 'bool', '*auto-dismiss': 'bool',
'*filter-node-name': 'str',
'*discard-source': 'bool',
log = iotests.filter_qemu_io(log)
return log
+ def do_cbw_error_via_blockdev_backup(self, on_cbw_error=None):
+ self.vm.cmd('blockdev-add', {
+ 'node-name': 'source',
+ 'driver': iotests.imgfmt,
+ 'file': {
+ 'driver': 'file',
+ 'filename': source_img
+ }
+ })
+
+ self.vm.cmd('blockdev-add', {
+ 'node-name': 'target',
+ 'driver': iotests.imgfmt,
+ 'file': {
+ 'driver': 'blkdebug',
+ 'image': {
+ 'driver': 'file',
+ 'filename': temp_img
+ },
+ 'inject-error': [
+ {
+ 'event': 'write_aio',
+ 'errno': 5,
+ 'immediately': False,
+ 'once': True
+ }
+ ]
+ }
+ })
+
+ blockdev_backup_options = {
+ 'device': 'source',
+ 'target': 'target',
+ 'sync': 'none',
+ 'job-id': 'job-id',
+ 'filter-node-name': 'cbw'
+ }
+
+ if on_cbw_error:
+ blockdev_backup_options['on-cbw-error'] = on_cbw_error
+
+ self.vm.cmd('blockdev-backup', blockdev_backup_options)
+
+ self.vm.cmd('blockdev-add', {
+ 'node-name': 'access',
+ 'driver': 'snapshot-access',
+ 'file': 'cbw'
+ })
+
+ result = self.vm.qmp('human-monitor-command',
+ command_line='qemu-io cbw "write 0 1M"')
+ self.assert_qmp(result, 'return', '')
+
+ result = self.vm.qmp('human-monitor-command',
+ command_line='qemu-io access "read 0 1M"')
+ self.assert_qmp(result, 'return', '')
+
+ self.vm.shutdown()
+ log = self.vm.get_log()
+ log = iotests.filter_qemu_io(log)
+ return log
+
def test_break_snapshot_on_cbw_error(self):
"""break-snapshot behavior:
Guest write succeed, but further snapshot-read fails, as snapshot is
write failed: Input/output error
read 1048576/1048576 bytes at offset 0
1 MiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
+""")
+
+ def test_break_snapshot_policy_forwarding(self):
+ """Ensure CBW filter accepts break-snapshot policy
+ specified in blockdev-backup QMP command.
+ """
+ log = self.do_cbw_error_via_blockdev_backup('break-snapshot')
+ self.assertEqual(log, """\
+wrote 1048576/1048576 bytes at offset 0
+1 MiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
+read failed: Permission denied
+""")
+
+ def test_break_guest_write_policy_forwarding(self):
+ """Ensure CBW filter accepts break-guest-write policy
+ specified in blockdev-backup QMP command.
+ """
+ log = self.do_cbw_error_via_blockdev_backup('break-guest-write')
+ self.assertEqual(log, """\
+write failed: Input/output error
+read 1048576/1048576 bytes at offset 0
+1 MiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
+""")
+
+ def test_default_on_cbw_error_policy_forwarding(self):
+ """Ensure break-guest-write policy is used by default when
+ on-cbw-error is not explicitly specified.
+ """
+ log = self.do_cbw_error_via_blockdev_backup()
+ self.assertEqual(log, """\
+write failed: Input/output error
+read 1048576/1048576 bytes at offset 0
+1 MiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
""")
def do_cbw_timeout(self, on_cbw_error):
-....
+.......
----------------------------------------------------------------------
-Ran 4 tests
+Ran 7 tests
OK
projects / thirdparty / qemu.git / commitdiff
