Apache
/
hbase
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

HBASE-19850 The number of Offline Regions is wrong after restoring a snapshot 

Signed-off-by: tedyu <yuzhihong@gmail.com>
This commit is contained in:
Toshihiro Suzuki 2018-01-24 03:34:45 +09:00 committed by tedyu
parent ce4c243399
commit 03eb3d24b6
2 changed files with 58 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

35
hbase-server/src/main/java/org/apache/hadoop/hbase/master/snapshot/RestoreSnapshotHandler.java
View File

@ -34,6 +34,7 @@ import org.apache.hadoop.hbase.HTableDescriptor;
import org.apache.hadoop.hbase.TableName;
import org.apache.hadoop.hbase.MetaTableAccessor;
import org.apache.hadoop.hbase.client.Connection;
import org.apache.hadoop.hbase.client.RegionReplicaUtil;
import org.apache.hadoop.hbase.errorhandling.ForeignException;
import org.apache.hadoop.hbase.errorhandling.ForeignExceptionDispatcher;
import org.apache.hadoop.hbase.executor.EventType;
@ -154,6 +155,9 @@ public class RestoreSnapshotHandler extends TableEventHandler implements Snapsho
if (metaChanges.hasRegionsToRemove()) hrisToRemove.addAll(metaChanges.getRegionsToRemove());
MetaTableAccessor.deleteRegions(conn, hrisToRemove);
// We also need to remove the current set of regions from in memory states
deleteRegionsFromInMemoryStates(hrisToRemove, hTableDescriptor.getRegionReplication());
// 4.2 Add the new set of regions to META
//
// At this point the old regions are no longer present in META.
@ -175,7 +179,6 @@ public class RestoreSnapshotHandler extends TableEventHandler implements Snapsho
RestoreSnapshotHelper.restoreSnapshotACL(snapshot, tableName, server.getConfiguration());
}
// At this point the restore is complete. Next step is enabling the table.
LOG.info("Restore snapshot=" + ClientSnapshotDescriptionUtils.toString(snapshot) +
" on table=" + tableName + " completed!");
@ -205,6 +208,36 @@ public class RestoreSnapshotHandler extends TableEventHandler implements Snapsho
}
}
/**
* Delete regions from in-memory states
* @param regionInfos regions to delete
* @param regionReplication the number of region replications
*/
private void deleteRegionsFromInMemoryStates(List<HRegionInfo> regionInfos,
int regionReplication) {
// Delete the regions from AssignmentManager
for (HRegionInfo hri : regionInfos) {
masterServices.getAssignmentManager().getRegionStates().deleteRegion(hri);
}
// Delete the regions from ServerManager
masterServices.getServerManager().removeRegions(regionInfos);
// For region replicas
if (regionReplication > 1) {
for (HRegionInfo regionInfo : regionInfos) {
for (int i = 1; i < regionReplication; i++) {
HRegionInfo regionInfoForReplica =
RegionReplicaUtil.getRegionInfoForReplica(regionInfo, i);
// Delete the regions from AssignmentManager
masterServices.getAssignmentManager().getRegionStates()
.deleteRegion(regionInfoForReplica);
// Delete the regions from ServerManager
masterServices.getServerManager().removeRegion(regionInfoForReplica);
}
}
}
}
@Override
protected void completed(final Throwable exception) {
this.stopped = true;

27
hbase-server/src/test/java/org/apache/hadoop/hbase/client/TestRestoreSnapshotFromClient.java
View File

@ -39,6 +39,7 @@ import org.apache.hadoop.hbase.MetaTableAccessor;
import org.apache.hadoop.hbase.coprocessor.BaseRegionObserver;
import org.apache.hadoop.hbase.coprocessor.ObserverContext;
import org.apache.hadoop.hbase.coprocessor.RegionCoprocessorEnvironment;
import org.apache.hadoop.hbase.master.RegionState;
import org.apache.hadoop.hbase.regionserver.InternalScanner;
import org.apache.hadoop.hbase.regionserver.ScanType;
import org.apache.hadoop.hbase.regionserver.Store;
@ -47,7 +48,6 @@ import org.apache.hadoop.hbase.TableName;
import org.apache.hadoop.hbase.Waiter;
import org.apache.hadoop.hbase.master.AssignmentManager;
import org.apache.hadoop.hbase.master.MasterFileSystem;
import org.apache.hadoop.hbase.master.RegionState;
import org.apache.hadoop.hbase.master.snapshot.SnapshotManager;
import org.apache.hadoop.hbase.regionserver.NoSuchColumnFamilyException;
import org.apache.hadoop.hbase.snapshot.CorruptedSnapshotException;
@ -384,8 +384,7 @@ public class TestRestoreSnapshotFromClient {
}
public static class DelayCompactionObserver extends BaseRegionObserver {
@Override
public InternalScanner preCompact(ObserverContext<RegionCoprocessorEnvironment> e,
@Override public InternalScanner preCompact(ObserverContext<RegionCoprocessorEnvironment> e,
final Store store, final InternalScanner scanner, final ScanType scanType)
throws IOException {
@ -400,6 +399,28 @@ public class TestRestoreSnapshotFromClient {
}
}
@Test
public void testOfflineRegionsShouldBeZeroAfterRestoreSnapshot() throws IOException,
InterruptedException {
// Load more data to split regions
SnapshotTestingUtils.loadData(TEST_UTIL, tableName, 2000, FAMILY);
// Split regions
List<HRegionInfo> regions = admin.getTableRegions(tableName);
RegionReplicaUtil.removeNonDefaultRegions(regions);
splitRegion(regions.get(0));
// Restore the snapshot
admin.disableTable(tableName);
admin.restoreSnapshot(snapshotName0);
admin.enableTable(tableName);
int offlineRegions = TEST_UTIL.getHBaseCluster().getMaster().getAssignmentManager()
.getRegionStates().getRegionByStateOfTable(tableName).get(RegionState.State.OFFLINE).size();
assertEquals(0, offlineRegions);
}
// ==========================================================================
// Helpers
// ==========================================================================