Skip to content

Commit a333735

Browse files
morozovThorneANN
authored andcommitted
[FLINK-38334][mysql] Fix MySQL CDC source stuck in INITIAL_ASSIGNING (apache#4278)
When a table is excluded from configuration after a restart from savepoint, the MySQL CDC source could get stuck in the INITIAL_ASSIGNING state. This happened because table exclusion cleanup was only performed when isAssigningFinished() was true, but the assigner couldn't finish because excluded table splits were never reported as finished. The fix separates two concerns in captureNewlyAddedTables(): - Adding new tables: should only happen when isAssigningFinished() - Removing excluded tables: must happen regardless of assigner status Added integration test TableExclusionDuringSnapshotIT that reproduces the issue by using a blocking hook to take a savepoint during INITIAL_ASSIGNING phase, then restarting with a table excluded from configuration.
1 parent b3c7c30 commit a333735

3 files changed

Lines changed: 319 additions & 2 deletions

File tree

flink-cdc-connect/flink-cdc-source-connectors/flink-connector-mysql-cdc/src/main/java/org/apache/flink/cdc/connectors/mysql/source/assigners/MySqlSnapshotSplitAssigner.java

Lines changed: 10 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -234,8 +234,7 @@ private void captureNewlyAddedTables() {
234234

235235
// Don't scan newly added table in snapshot mode.
236236
if (sourceConfig.isScanNewlyAddedTableEnabled()
237-
&& !sourceConfig.getStartupOptions().isSnapshotOnly()
238-
&& AssignerStatus.isAssigningFinished(assignerStatus)) {
237+
&& !sourceConfig.getStartupOptions().isSnapshotOnly()) {
239238
// check whether we got newly added tables
240239
try (JdbcConnection jdbc = DebeziumUtils.openJdbcConnection(sourceConfig)) {
241240
final List<TableId> currentCapturedTables =
@@ -258,6 +257,10 @@ private void captureNewlyAddedTables() {
258257
List<TableId> newlyAddedTables = currentCapturedTables;
259258

260259
// case 1: there are old tables to remove from state
260+
// Table removal must happen regardless of assigner status. When a table
261+
// is excluded after splits have been assigned but before they are finished,
262+
// we must remove those splits to prevent the assigner from waiting indefinitely
263+
// for splits that will never be reported as finished.
261264
if (!tablesToRemove.isEmpty()) {
262265

263266
// remove unassigned tables/splits if it does not satisfy new table filter
@@ -279,6 +282,11 @@ private void captureNewlyAddedTables() {
279282
alreadyProcessedTables.removeIf(tableId -> tablesToRemove.contains(tableId));
280283
}
281284

285+
// Adding new tables should only happen when assigning is finished.
286+
if (!AssignerStatus.isAssigningFinished(assignerStatus)) {
287+
return;
288+
}
289+
282290
// case 2: there are new tables to add
283291
if (!newlyAddedTables.isEmpty()) {
284292
LOG.info("Found newly added tables, start capture newly added tables process");
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,283 @@
1+
/*
2+
* Licensed to the Apache Software Foundation (ASF) under one or more
3+
* contributor license agreements. See the NOTICE file distributed with
4+
* this work for additional information regarding copyright ownership.
5+
* The ASF licenses this file to You under the Apache License, Version 2.0
6+
* (the "License"); you may not use this file except in compliance with
7+
* the License. You may obtain a copy of the License at
8+
*
9+
* http://www.apache.org/licenses/LICENSE-2.0
10+
*
11+
* Unless required by applicable law or agreed to in writing, software
12+
* distributed under the License is distributed on an "AS IS" BASIS,
13+
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
14+
* See the License for the specific language governing permissions and
15+
* limitations under the License.
16+
*/
17+
18+
package org.apache.flink.cdc.connectors.mysql.source;
19+
20+
import org.apache.flink.api.common.eventtime.WatermarkStrategy;
21+
import org.apache.flink.cdc.connectors.mysql.debezium.DebeziumUtils;
22+
import org.apache.flink.cdc.connectors.mysql.source.split.MySqlSnapshotSplit;
23+
import org.apache.flink.cdc.connectors.mysql.source.utils.hooks.SnapshotPhaseHooks;
24+
import org.apache.flink.cdc.connectors.mysql.testutils.UniqueDatabase;
25+
import org.apache.flink.cdc.debezium.table.RowDataDebeziumDeserializeSchema;
26+
import org.apache.flink.configuration.Configuration;
27+
import org.apache.flink.configuration.StateRecoveryOptions;
28+
import org.apache.flink.core.execution.CheckpointingMode;
29+
import org.apache.flink.core.execution.JobClient;
30+
import org.apache.flink.core.execution.SavepointFormatType;
31+
import org.apache.flink.runtime.checkpoint.CheckpointException;
32+
import org.apache.flink.runtime.dispatcher.UnavailableDispatcherOperationException;
33+
import org.apache.flink.streaming.api.datastream.DataStreamSource;
34+
import org.apache.flink.streaming.api.environment.StreamExecutionEnvironment;
35+
import org.apache.flink.table.api.DataTypes;
36+
import org.apache.flink.table.data.RowData;
37+
import org.apache.flink.table.data.conversion.RowRowConverter;
38+
import org.apache.flink.table.runtime.typeutils.InternalTypeInfo;
39+
import org.apache.flink.table.types.DataType;
40+
import org.apache.flink.table.types.logical.RowType;
41+
import org.apache.flink.table.types.utils.TypeConversions;
42+
import org.apache.flink.util.CloseableIterator;
43+
import org.apache.flink.util.ExceptionUtils;
44+
45+
import io.debezium.connector.mysql.MySqlConnection;
46+
import org.assertj.core.api.Assertions;
47+
import org.junit.jupiter.api.BeforeEach;
48+
import org.junit.jupiter.api.Test;
49+
import org.junit.jupiter.api.Timeout;
50+
import org.junit.jupiter.api.io.TempDir;
51+
52+
import java.nio.file.Path;
53+
import java.sql.SQLException;
54+
import java.time.ZoneId;
55+
import java.util.Arrays;
56+
import java.util.HashMap;
57+
import java.util.Map;
58+
import java.util.Optional;
59+
import java.util.Properties;
60+
import java.util.concurrent.CountDownLatch;
61+
62+
/**
63+
* IT test for FLINK-38334: MySQL CDC source gets stuck in INITIAL_ASSIGNING state when a table is
64+
* excluded from configuration after splits have been assigned but before they are finished.
65+
*
66+
* <p>NOTE: The test uses a JUnit timeout as a CI safeguard. The test logic is deterministic: it
67+
* either receives a binlog record (proving streaming mode) or blocks forever (bug exists).
68+
*
69+
* <p>When the bug exists, the test will timeout because:
70+
*
71+
* <ol>
72+
* <li>The savepoint captures table "a" splits in assignedSplits (assigned but not finished)
73+
* <li>After restart with table "a" excluded, the reader skips table "a" splits
74+
* <li>The enumerator waits for table "a" splits to be reported as finished (they never will be)
75+
* <li>allSnapshotSplitsFinished() never returns true → job stuck in INITIAL_ASSIGNING
76+
* </ol>
77+
*
78+
* <p>When the fix is applied, the test passes because excluded table splits are cleaned up on
79+
* restore, allowing the job to transition to streaming mode.
80+
*/
81+
public class TableExclusionDuringSnapshotIT extends MySqlSourceTestBase {
82+
private static final UniqueDatabase DATABASE =
83+
new UniqueDatabase(MYSQL_CONTAINER, "table_exclusion_snapshot", "mysqluser", "mysqlpw");
84+
private static final DataType DATA_TYPE = DataTypes.ROW(DataTypes.FIELD("id", DataTypes.INT()));
85+
private static final RowRowConverter ROW_CONVERTER = RowRowConverter.create(DATA_TYPE);
86+
private static final RowDataDebeziumDeserializeSchema DESERIALIZER =
87+
RowDataDebeziumDeserializeSchema.newBuilder()
88+
.setPhysicalRowType((RowType) DATA_TYPE.getLogicalType())
89+
.setResultTypeInfo(
90+
InternalTypeInfo.of(TypeConversions.fromDataToLogicalType(DATA_TYPE)))
91+
.build();
92+
93+
@BeforeEach
94+
void setUp() {
95+
DATABASE.createAndInitialize();
96+
}
97+
98+
// Latches for coordinating between test thread and snapshot hook
99+
// These are static because the hook is serialized and deserialized
100+
private static volatile CountDownLatch hookTriggeredLatch;
101+
private static volatile CountDownLatch savepointTakenLatch;
102+
103+
/**
104+
* Tests that excluding a table from configuration during INITIAL_ASSIGNING phase doesn't cause
105+
* the source to get stuck.
106+
*
107+
* <p>Scenario:
108+
*
109+
* <ol>
110+
* <li>Start job capturing tables "a" and "b" with a blocking hook on table "a"
111+
* <li>Take savepoint while table "a" is being snapshotted (splits assigned but not finished)
112+
* <li>Restart with configuration excluding table "a"
113+
* <li>Insert a new record into table "b"
114+
* <li>Verify we receive the new record (proves job transitioned to streaming)
115+
* </ol>
116+
*
117+
* <p>If the bug exists, the job will be stuck in INITIAL_ASSIGNING because the enumerator waits
118+
* for table "a" splits to be reported as finished, but the reader skips them.
119+
*/
120+
@Test
121+
@Timeout(120)
122+
void testTableExclusionDuringInitialAssigning(@TempDir Path tempDir) throws Exception {
123+
final String savepointDirectory = tempDir.toUri().toString();
124+
125+
executeSql("INSERT INTO a VALUES (1)");
126+
executeSql("INSERT INTO b VALUES (1)");
127+
executeSql("INSERT INTO b VALUES (2)");
128+
129+
// Phase 1: Take savepoint while table "a" splits are assigned but not finished
130+
String savepointPath =
131+
runJobAndSavepointDuringInitialAssigning(savepointDirectory, "a", "b");
132+
133+
// Phase 2: Restart with only table "b", verify streaming mode works
134+
String binlogRecord = runJobFromSavepointAndVerifyStreaming(savepointPath, "b");
135+
Assertions.assertThat(binlogRecord).isEqualTo("+I[200]");
136+
}
137+
138+
/** Starts a job and takes a savepoint while splits are assigned but not finished. */
139+
private String runJobAndSavepointDuringInitialAssigning(
140+
String savepointDirectory, String... tableNames) throws Exception {
141+
hookTriggeredLatch = new CountDownLatch(1);
142+
savepointTakenLatch = new CountDownLatch(1);
143+
144+
StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();
145+
env.setParallelism(1);
146+
env.enableCheckpointing(200, CheckpointingMode.EXACTLY_ONCE);
147+
148+
MySqlSource<RowData> source = createSourceWithBlockingHook(tableNames);
149+
DataStreamSource<RowData> stream =
150+
env.fromSource(source, WatermarkStrategy.noWatermarks(), "MySQL CDC Source");
151+
stream.print();
152+
153+
JobClient jobClient = env.executeAsync("Snapshot phase");
154+
155+
hookTriggeredLatch.await();
156+
String savepointPath = triggerSavepointWithRetry(jobClient, savepointDirectory);
157+
savepointTakenLatch.countDown();
158+
159+
jobClient.cancel().get();
160+
return savepointPath;
161+
}
162+
163+
/** Restarts job from savepoint and verifies it transitions to streaming mode. */
164+
private String runJobFromSavepointAndVerifyStreaming(String savepointPath, String... tableNames)
165+
throws Exception {
166+
Configuration configuration = new Configuration();
167+
configuration.set(StateRecoveryOptions.SAVEPOINT_PATH, savepointPath);
168+
169+
StreamExecutionEnvironment env =
170+
StreamExecutionEnvironment.getExecutionEnvironment(configuration);
171+
env.setParallelism(1);
172+
env.enableCheckpointing(500, CheckpointingMode.EXACTLY_ONCE);
173+
174+
MySqlSource<RowData> source = createSourceBuilder(tableNames).build();
175+
DataStreamSource<RowData> stream =
176+
env.fromSource(source, WatermarkStrategy.noWatermarks(), "MySQL CDC Source");
177+
178+
try (CloseableIterator<RowData> iterator = stream.executeAndCollect()) {
179+
// Consume snapshot records from table "b" (2 rows)
180+
for (int i = 0; i < 2; i++) {
181+
iterator.next();
182+
}
183+
184+
// Insert a new record - if job is in streaming mode, we'll receive it
185+
executeSql("INSERT INTO b VALUES (200)");
186+
187+
// This blocks forever if the job is stuck in INITIAL_ASSIGNING
188+
return ROW_CONVERTER.toExternal(iterator.next()).toString();
189+
}
190+
}
191+
192+
private MySqlSourceBuilder<RowData> createSourceBuilder(String... tableNames) {
193+
return MySqlSource.<RowData>builder()
194+
.hostname(MYSQL_CONTAINER.getHost())
195+
.port(MYSQL_CONTAINER.getDatabasePort())
196+
.databaseList(DATABASE.getDatabaseName())
197+
.serverTimeZone("UTC")
198+
.tableList(
199+
Arrays.stream(tableNames)
200+
.map(t -> DATABASE.getDatabaseName() + "." + t)
201+
.toArray(String[]::new))
202+
.username(DATABASE.getUsername())
203+
.password(DATABASE.getPassword())
204+
.deserializer(DESERIALIZER)
205+
.scanNewlyAddedTableEnabled(true);
206+
}
207+
208+
/** Creates a source with a hook that blocks on table "a" until the savepoint is taken. */
209+
private MySqlSource<RowData> createSourceWithBlockingHook(String... tableNames) {
210+
MySqlSource<RowData> source = createSourceBuilder(tableNames).build();
211+
212+
SnapshotPhaseHooks hooks = new SnapshotPhaseHooks();
213+
hooks.setPostLowWatermarkAction(
214+
(connection, split) -> {
215+
MySqlSnapshotSplit snapshotSplit = (MySqlSnapshotSplit) split;
216+
if (!snapshotSplit.getTableId().table().equals("a")) {
217+
return;
218+
}
219+
hookTriggeredLatch.countDown();
220+
try {
221+
savepointTakenLatch.await();
222+
} catch (InterruptedException e) {
223+
Thread.currentThread().interrupt();
224+
}
225+
});
226+
source.setSnapshotHooks(hooks);
227+
228+
return source;
229+
}
230+
231+
private String triggerSavepointWithRetry(JobClient jobClient, String savepointDirectory)
232+
throws Exception {
233+
int retryTimes = 0;
234+
while (retryTimes < 600) {
235+
try {
236+
return jobClient
237+
.triggerSavepoint(savepointDirectory, SavepointFormatType.DEFAULT)
238+
.get();
239+
} catch (Exception e) {
240+
// Retry if checkpoint triggering task is not yet ready
241+
Optional<CheckpointException> checkpointException =
242+
ExceptionUtils.findThrowable(e, CheckpointException.class);
243+
if (checkpointException.isPresent()
244+
&& checkpointException
245+
.get()
246+
.getMessage()
247+
.contains("Checkpoint triggering task")) {
248+
Thread.sleep(100);
249+
retryTimes++;
250+
continue;
251+
}
252+
// Retry if job is still initializing
253+
Optional<UnavailableDispatcherOperationException> dispatcherException =
254+
ExceptionUtils.findThrowable(
255+
e, UnavailableDispatcherOperationException.class);
256+
if (dispatcherException.isPresent()) {
257+
Thread.sleep(100);
258+
retryTimes++;
259+
continue;
260+
}
261+
throw e;
262+
}
263+
}
264+
throw new RuntimeException("Failed to trigger savepoint after " + retryTimes + " retries");
265+
}
266+
267+
private void executeSql(String... statements) throws SQLException {
268+
Map<String, String> properties = new HashMap<>();
269+
properties.put("database.hostname", MYSQL_CONTAINER.getHost());
270+
properties.put("database.port", String.valueOf(MYSQL_CONTAINER.getDatabasePort()));
271+
properties.put("database.user", DATABASE.getUsername());
272+
properties.put("database.password", DATABASE.getPassword());
273+
properties.put("database.serverTimezone", ZoneId.of("UTC").toString());
274+
io.debezium.config.Configuration configuration =
275+
io.debezium.config.Configuration.from(properties);
276+
277+
try (MySqlConnection connection =
278+
DebeziumUtils.createMySqlConnection(configuration, new Properties())) {
279+
connection.execute("USE " + DATABASE.getDatabaseName());
280+
connection.execute(statements);
281+
}
282+
}
283+
}
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,26 @@
1+
-- Licensed to the Apache Software Foundation (ASF) under one or more
2+
-- contributor license agreements. See the NOTICE file distributed with
3+
-- this work for additional information regarding copyright ownership.
4+
-- The ASF licenses this file to You under the Apache License, Version 2.0
5+
-- (the "License"); you may not use this file except in compliance with
6+
-- the License. You may obtain a copy of the License at
7+
--
8+
-- http://www.apache.org/licenses/LICENSE-2.0
9+
--
10+
-- Unless required by applicable law or agreed to in writing, software
11+
-- distributed under the License is distributed on an "AS IS" BASIS,
12+
-- WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
13+
-- See the License for the specific language governing permissions and
14+
-- limitations under the License.
15+
16+
-- DATABASE: table_exclusion_snapshot
17+
CREATE TABLE a (
18+
id INT NOT NULL,
19+
PRIMARY KEY (id)
20+
);
21+
22+
CREATE TABLE b (
23+
id INT NOT NULL,
24+
PRIMARY KEY (id)
25+
);
26+

0 commit comments

Comments
 (0)