[
https://issues.apache.org/jira/browse/FLINK-8790?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16497753#comment-16497753
]
ASF GitHub Bot commented on FLINK-8790:
---------------------------------------
Github user sihuazhou commented on a diff in the pull request:
https://github.com/apache/flink/pull/5582#discussion_r192335886
--- Diff:
flink-state-backends/flink-statebackend-rocksdb/src/main/java/org/apache/flink/contrib/streaming/state/RocksDBKeyedStateBackend.java
---
@@ -795,20 +806,241 @@ private void restoreInstance(
}
/**
- * Recovery from local incremental state.
+ * Recovery from multi incremental states.
+ * In case of rescaling, this method creates a temporary
RocksDB instance for a key-groups shard. All contents
+ * from the temporary instance are copied into the real restore
instance and then the temporary instance is
+ * discarded.
*/
- private void restoreInstance(IncrementalLocalKeyedStateHandle
localKeyedStateHandle) throws Exception {
+ void restoreFromMultiHandles(Collection<KeyedStateHandle>
restoreStateHandles) throws Exception {
+
+ KeyGroupRange targetKeyGroupRange =
stateBackend.keyGroupRange;
+
+ chooseTheBestStateHandleToInit(restoreStateHandles,
targetKeyGroupRange);
+
+ int targetStartKeyGroup =
stateBackend.getKeyGroupRange().getStartKeyGroup();
+ byte[] targetStartKeyGroupPrefixBytes = new
byte[stateBackend.keyGroupPrefixBytes];
+ for (int j = 0; j < stateBackend.keyGroupPrefixBytes;
++j) {
+ targetStartKeyGroupPrefixBytes[j] = (byte)
(targetStartKeyGroup >>> ((stateBackend.keyGroupPrefixBytes - j - 1) *
Byte.SIZE));
+ }
+
+ for (KeyedStateHandle rawStateHandle :
restoreStateHandles) {
+
+ if (!(rawStateHandle instanceof
IncrementalKeyedStateHandle)) {
+ throw new
IllegalStateException("Unexpected state handle type, " +
+ "expected " +
IncrementalKeyedStateHandle.class +
+ ", but found " +
rawStateHandle.getClass());
+ }
+
+ Path temporaryRestoreInstancePath = new
Path(stateBackend.instanceBasePath.getAbsolutePath() +
UUID.randomUUID().toString());
+ try (RestoredDBInstance tmpRestoreDBInfo =
restoreDBFromStateHandle(
+ (IncrementalKeyedStateHandle)
rawStateHandle,
+ temporaryRestoreInstancePath,
+ targetKeyGroupRange,
+
stateBackend.keyGroupPrefixBytes,
+ false);
+ RocksDBWriteBatchWrapper
writeBatchWrapper = new RocksDBWriteBatchWrapper(stateBackend.db)) {
+
+ List<ColumnFamilyDescriptor>
tmpColumnFamilyDescriptors = tmpRestoreDBInfo.columnFamilyDescriptors;
+ List<ColumnFamilyHandle>
tmpColumnFamilyHandles = tmpRestoreDBInfo.columnFamilyHandles;
+
+ int startKeyGroup =
stateBackend.getKeyGroupRange().getStartKeyGroup();
+ byte[] startKeyGroupPrefixBytes = new
byte[stateBackend.keyGroupPrefixBytes];
+ for (int j = 0; j <
stateBackend.keyGroupPrefixBytes; ++j) {
+ startKeyGroupPrefixBytes[j] =
(byte) (startKeyGroup >>> ((stateBackend.keyGroupPrefixBytes - j - 1) *
Byte.SIZE));
+ }
+
+ // iterating only the requested
descriptors automatically skips the default column family handle
+ for (int i = 0; i <
tmpColumnFamilyDescriptors.size(); ++i) {
+ ColumnFamilyHandle
tmpColumnFamilyHandle = tmpColumnFamilyHandles.get(i);
+ ColumnFamilyDescriptor
tmpColumnFamilyDescriptor = tmpColumnFamilyDescriptors.get(i);
+
+ ColumnFamilyHandle
targetColumnFamilyHandle = getOrRegisterColumnFamilyHandle(
+
tmpColumnFamilyDescriptor, null,
tmpRestoreDBInfo.stateMetaInfoSnapshots.get(i));
+
+ try (RocksIteratorWrapper
iterator = getRocksIterator(tmpRestoreDBInfo.db, tmpColumnFamilyHandle)) {
+
+
iterator.seek(startKeyGroupPrefixBytes);
+
+ while
(iterator.isValid()) {
+
+ int keyGroup =
0;
+ for (int j = 0;
j < stateBackend.keyGroupPrefixBytes; ++j) {
+
keyGroup = (keyGroup << Byte.SIZE) + iterator.key()[j];
+ }
+
+ if
(stateBackend.keyGroupRange.contains(keyGroup)) {
+
writeBatchWrapper.put(targetColumnFamilyHandle, iterator.key(),
iterator.value());
+ } else {
+ //
Since the iterator will visit the record according to the sorted order,
+ // we
can just break here.
+ break;
+ }
+
+ iterator.next();
+ }
+ } // releases native iterator
resources
+ }
+ } finally {
+ FileSystem restoreFileSystem =
temporaryRestoreInstancePath.getFileSystem();
+ if
(restoreFileSystem.exists(temporaryRestoreInstancePath)) {
+
restoreFileSystem.delete(temporaryRestoreInstancePath, true);
+ }
+ }
+ }
+ }
+
+ private class RestoredDBInstance implements AutoCloseable {
+
+ @Nonnull
+ private final RocksDB db;
+
+ @Nonnull
+ private final ColumnFamilyHandle
defaultColumnFamilyHandle;
+
+ @Nonnull
+ private final List<ColumnFamilyHandle>
columnFamilyHandles;
+
+ @Nonnull
+ private final List<ColumnFamilyDescriptor>
columnFamilyDescriptors;
+
+ @Nonnull
+ private final
List<RegisteredKeyedBackendStateMetaInfo.Snapshot<?, ?>> stateMetaInfoSnapshots;
+
+ public RestoredDBInstance(@Nonnull RocksDB db,
+ @Nonnull
List<ColumnFamilyHandle> columnFamilyHandles,
+ @Nonnull
List<ColumnFamilyDescriptor> columnFamilyDescriptors,
+ @Nonnull
List<RegisteredKeyedBackendStateMetaInfo.Snapshot<?, ?>>
stateMetaInfoSnapshots) {
+ this.db = db;
+ this.columnFamilyHandles = columnFamilyHandles;
+ this.defaultColumnFamilyHandle =
this.columnFamilyHandles.remove(0);
+ this.columnFamilyDescriptors =
columnFamilyDescriptors;
+ this.stateMetaInfoSnapshots =
stateMetaInfoSnapshots;
+ }
+
+ @Override
+ public void close() throws Exception {
--- End diff --
👍
> Improve performance for recovery from incremental checkpoint
> ------------------------------------------------------------
>
> Key: FLINK-8790
> URL: https://issues.apache.org/jira/browse/FLINK-8790
> Project: Flink
> Issue Type: Improvement
> Components: State Backends, Checkpointing
> Affects Versions: 1.5.0
> Reporter: Sihua Zhou
> Assignee: Sihua Zhou
> Priority: Major
> Fix For: 1.6.0
>
>
> When there are multi state handle to be restored, we can improve the
> performance as follow:
> 1. Choose the best state handle to init the target db
> 2. Use the other state handles to create temp db, and clip the db according
> to the target key group range (via rocksdb.deleteRange()), this can help use
> get rid of the `key group check` in
> `data insertion loop` and also help us get rid of traversing the useless
> record.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
