You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@flink.apache.org by GitBox <gi...@apache.org> on 2022/07/01 14:53:16 UTC
[GitHub] [flink] dzikosc opened a new pull request, #20133: Kinesis Consumer stop-with-savepoint support including EFO
dzikosc opened a new pull request, #20133:
URL: https://github.com/apache/flink/pull/20133
<!--
*Thank you very much for contributing to Apache Flink - we are happy that you want to help us improve Flink. To help the community review your contribution in the best possible way, please go through the checklist below, which will get the contribution into a shape in which it can be best reviewed.*
*Please understand that we do not do this to make contributions to Flink a hassle. In order to uphold a high standard of quality for code contributions, while at the same time managing a large number of contributions, we need contributors to prepare the contributions well, and give reviewers enough contextual information for the review. Please also understand that contributions that do not follow this guide will take longer to review and thus typically be picked up with lower priority by the community.*
## Contribution Checklist
- Make sure that the pull request corresponds to a [JIRA issue](https://issues.apache.org/jira/projects/FLINK/issues). Exceptions are made for typos in JavaDoc or documentation files, which need no JIRA issue.
- Name the pull request in the form "[FLINK-XXXX] [component] Title of the pull request", where *FLINK-XXXX* should be replaced by the actual issue number. Skip *component* if you are unsure about which is the best component.
Typo fixes that have no associated JIRA issue should be named following this pattern: `[hotfix] [docs] Fix typo in event time introduction` or `[hotfix] [javadocs] Expand JavaDoc for PuncuatedWatermarkGenerator`.
- Fill out the template below to describe the changes contributed by the pull request. That will give reviewers the context they need to do the review.
- Make sure that the change passes the automated tests, i.e., `mvn clean verify` passes. You can set up Azure Pipelines CI to do that following [this guide](https://cwiki.apache.org/confluence/display/FLINK/Azure+Pipelines#AzurePipelines-Tutorial:SettingupAzurePipelinesforaforkoftheFlinkrepository).
- Each pull request should address only one issue, not mix up code from multiple issues.
- Each commit in the pull request has a meaningful commit message (including the JIRA id)
- Once all items of the checklist are addressed, remove the above text and this checklist, leaving only the filled out template below.
**(The sections below can be removed for hotfixes of typos)**
-->
## What is the purpose of the change
*(For example: This pull request makes task deployment go through the blob server, rather than through RPC. That way we avoid re-transferring them on each deployment (during recovery).)*
## Brief change log
*(for example:)*
- *The TaskInfo is stored in the blob store on job creation time as a persistent artifact*
- *Deployments RPC transmits only the blob storage reference*
- *TaskManagers retrieve the TaskInfo from the blob cache*
## Verifying this change
Please make sure both new and modified tests in this PR follows the conventions defined in our code quality guide: https://flink.apache.org/contributing/code-style-and-quality-common.html#testing
*(Please pick either of the following options)*
This change is a trivial rework / code cleanup without any test coverage.
*(or)*
This change is already covered by existing tests, such as *(please describe tests)*.
*(or)*
This change added tests and can be verified as follows:
*(example:)*
- *Added integration tests for end-to-end deployment with large payloads (100MB)*
- *Extended integration test for recovery after master (JobManager) failure*
- *Added test that validates that TaskInfo is transferred only once across recoveries*
- *Manually verified the change by running a 4 node cluster with 2 JobManagers and 4 TaskManagers, a stateful streaming program, and killing one JobManager and two TaskManagers during the execution, verifying that recovery happens correctly.*
## Does this pull request potentially affect one of the following parts:
- Dependencies (does it add or upgrade a dependency): (yes / no)
- The public API, i.e., is any changed class annotated with `@Public(Evolving)`: (yes / no)
- The serializers: (yes / no / don't know)
- The runtime per-record code paths (performance sensitive): (yes / no / don't know)
- Anything that affects deployment or recovery: JobManager (and its components), Checkpointing, Kubernetes/Yarn, ZooKeeper: (yes / no / don't know)
- The S3 file system connector: (yes / no / don't know)
## Documentation
- Does this pull request introduce a new feature? (yes / no)
- If yes, how is the feature documented? (not applicable / docs / JavaDocs / not documented)
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: issues-unsubscribe@flink.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [flink] dzikosc commented on a diff in pull request #20133: Kinesis Consumer stop-with-savepoint support including EFO
Posted by GitBox <gi...@apache.org>.
dzikosc commented on code in PR #20133:
URL: https://github.com/apache/flink/pull/20133#discussion_r913039077
##########
flink-connectors/flink-connector-kinesis/src/test/java/org/apache/flink/streaming/connectors/kinesis/FlinkKinesisITCase.java:
##########
@@ -69,104 +81,174 @@ public class FlinkKinesisITCase extends TestLogger {
@Rule public TemporaryFolder temp = new TemporaryFolder();
+ @Rule public SharedObjects sharedObjects = SharedObjects.create();
+
private static final SimpleStringSchema STRING_SCHEMA = new SimpleStringSchema();
private KinesisPubsubClient client;
@Before
- public void setupClient() {
- client = new KinesisPubsubClient(kinesalite.getContainerProperties());
+ public void setupClient() throws Exception {
+ client = new KinesisPubsubClient(getContainerProperties());
+ stream = TestNameProvider.getCurrentTestName().replaceAll("\\W", "");
+ client.createTopic(stream, 1, new Properties());
+ }
+
+ @Test
+ public void testStopWithSavepoint() throws Exception {
+ testStopWithSavepoint(false, false);
+ }
+
+ @Test
+ public void testStopWithSavepointWithDrain() throws Exception {
+ testStopWithSavepoint(true, false);
+ }
+
+ @Test
+ @Ignore("Kinesalite does not support EFO")
+ public void testStopWithSavepointWithEfo() throws Exception {
+ testStopWithSavepoint(false, true);
+ }
+
+ @Test
+ @Ignore("Kinesalite does not support EFO")
+ public void testStopWithSavepointWithDrainAndEfo() throws Exception {
+ testStopWithSavepoint(true, true);
}
/**
* Tests that pending elements do not cause a deadlock during stop with savepoint (FLINK-17170).
*
* <ol>
- * <li>The test setups up a stream with 100 records and creates a Flink job that reads them
+ * <li>The test setups up a stream with 1000 records and creates a Flink job that reads them
* with very slowly (using up a large chunk of time of the mailbox).
* <li>After ensuring that consumption has started, the job is stopped in a parallel thread.
* <li>Without the fix of FLINK-17170, the job now has a high chance to deadlock during
* cancel.
* <li>With the fix, the job proceeds and we can lift the backpressure.
* </ol>
*/
- @Test
- public void testStopWithSavepoint() throws Exception {
- client.createTopic(TEST_STREAM, 1, new Properties());
-
+ private void testStopWithSavepoint(boolean drain, boolean efo) throws Exception {
// add elements to the test stream
int numElements = 1000;
client.sendMessage(
- TEST_STREAM,
+ stream,
IntStream.range(0, numElements).mapToObj(String::valueOf).toArray(String[]::new));
StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();
env.setParallelism(1);
+ env.enableCheckpointing(100L);
- Properties config = kinesalite.getContainerProperties();
- config.setProperty(STREAM_INITIAL_POSITION, InitialPosition.TRIM_HORIZON.name());
- FlinkKinesisConsumer<String> consumer =
- new FlinkKinesisConsumer<>(TEST_STREAM, STRING_SCHEMA, config);
+ SharedReference<CountDownLatch> savepointTrigger = sharedObjects.add(new CountDownLatch(1));
+ DataStream<String> outputStream =
+ env.addSource(createKinesisConsumer(efo)).map(new WaitingMapper(savepointTrigger));
- DataStream<String> stream = env.addSource(consumer).map(new WaitingMapper());
// call stop with savepoint in another thread
ForkJoinTask<Object> stopTask =
ForkJoinPool.commonPool()
.submit(
() -> {
- WaitingMapper.firstElement.await();
- stopWithSavepoint();
- WaitingMapper.stopped = true;
+ savepointTrigger.get().await();
+ stopWithSavepoint(drain);
return null;
});
try {
- List<String> result = stream.executeAndCollect(10000);
+ List<String> result = outputStream.executeAndCollect(10000);
// stop with savepoint will most likely only return a small subset of the elements
// validate that the prefix is as expected
- assertThat(result, hasSize(lessThan(numElements)));
- assertThat(
- result,
- equalTo(
- IntStream.range(0, numElements)
- .mapToObj(String::valueOf)
- .collect(Collectors.toList())
- .subList(0, result.size())));
+ if (drain) {
+ assertThat(
+ result,
+ contains(
+ IntStream.range(0, numElements)
+ .mapToObj(String::valueOf)
+ .toArray()));
+ } else {
+ // stop with savepoint will most likely only return a small subset of the elements
+ // validate that the prefix is as expected
+ assertThat(
+ result,
+ contains(
+ IntStream.range(0, result.size())
+ .mapToObj(String::valueOf)
+ .toArray()));
+ }
} finally {
- stopTask.cancel(true);
+ stopTask.get();
+ }
+ }
+
+ private FlinkKinesisConsumer<String> createKinesisConsumer(boolean efo) {
+ Properties config = getContainerProperties();
+ config.setProperty(STREAM_INITIAL_POSITION, InitialPosition.TRIM_HORIZON.name());
+ if (efo) {
+ config.putIfAbsent(RECORD_PUBLISHER_TYPE, "EFO");
+ config.putIfAbsent(EFO_CONSUMER_NAME, "efo-flink-app");
}
+ return new FlinkKinesisConsumer<>(stream, STRING_SCHEMA, config);
+ }
+
+ private Properties getContainerProperties() {
+ return kinesalite.getContainerProperties();
}
- private String stopWithSavepoint() throws Exception {
+ private String stopWithSavepoint(boolean drain) throws Exception {
JobStatusMessage job =
MINI_CLUSTER.getClusterClient().listJobs().get().stream().findFirst().get();
return MINI_CLUSTER
.getClusterClient()
.stopWithSavepoint(
job.getJobId(),
- true,
+ drain,
temp.getRoot().getAbsolutePath(),
SavepointFormatType.CANONICAL)
.get();
}
- private static class WaitingMapper implements MapFunction<String, String> {
- static CountDownLatch firstElement;
- static volatile boolean stopped;
+ private static class WaitingMapper
+ implements MapFunction<String, String>, CheckpointedFunction {
+ private final SharedReference<CountDownLatch> savepointTrigger;
+ private volatile boolean savepointTriggered;
+ // keep track on when the last checkpoint occurred
+ private transient Deadline checkpointDeadline;
+ private final AtomicInteger numElements = new AtomicInteger();
+
+ WaitingMapper(SharedReference<CountDownLatch> savepointTrigger) {
+ this.savepointTrigger = savepointTrigger;
+ checkpointDeadline = Deadline.fromNow(Duration.ofDays(1));
+ }
- WaitingMapper() {
- firstElement = new CountDownLatch(1);
- stopped = false;
+ private void readObject(ObjectInputStream stream)
+ throws ClassNotFoundException, IOException {
+ stream.defaultReadObject();
+ checkpointDeadline = Deadline.fromNow(Duration.ofDays(1));
}
@Override
public String map(String value) throws Exception {
- if (firstElement.getCount() > 0) {
- firstElement.countDown();
- }
- if (!stopped) {
- Thread.sleep(100);
+ numElements.incrementAndGet();
+ if (!savepointTriggered) {
+ try {
+ Thread.sleep(100);
+ } catch (InterruptedException e) {
Review Comment:
The primary function of this mapper is to slow down data ingestion until savepoint is triggered.
We could mark the operator thread as interrupted, but I don't think this would have any broader impact.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: issues-unsubscribe@flink.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [flink] dannycranmer commented on a diff in pull request #20133: Kinesis Consumer stop-with-savepoint support including EFO
Posted by GitBox <gi...@apache.org>.
dannycranmer commented on code in PR #20133:
URL: https://github.com/apache/flink/pull/20133#discussion_r912129760
##########
flink-connectors/flink-connector-kinesis/src/main/java/org/apache/flink/streaming/connectors/kinesis/internals/publisher/fanout/FanOutRecordPublisher.java:
##########
@@ -154,7 +155,8 @@ private RecordPublisherRunResult runWithBackoff(
fanOutShardSubscriber.subscribeToShardAndConsumeRecords(
toSdkV2StartingPosition(nextStartingPosition), eventConsumer);
attempt = 0;
- } catch (FanOutSubscriberInterruptedException ex) {
+ } catch (FanOutSubscriberInterruptedException | RejectedExecutionException ex) {
+ // RejectedExecutionException occurs during soft shutdown, e.g. stop with savepoint
Review Comment:
Can you please add a unit test for this change?
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: issues-unsubscribe@flink.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [flink] dannycranmer commented on a diff in pull request #20133: [FLINK-23528][connectors/kinesis] Kinesis Consumer stop-with-savepoint support including EFO
Posted by GitBox <gi...@apache.org>.
dannycranmer commented on code in PR #20133:
URL: https://github.com/apache/flink/pull/20133#discussion_r927497905
##########
flink-connectors/flink-connector-kinesis/src/main/java/org/apache/flink/streaming/connectors/kinesis/internals/KinesisDataFetcher.java:
##########
@@ -852,6 +852,15 @@ protected void deregisterStreamConsumer() {
StreamConsumerRegistrarUtil.deregisterStreamConsumers(configProps, streams);
}
+ /**
+ * Gracefully stops shardConsumersExecutor without interrupting running threads. Allows test to
+ * override this to simulate exception for shutdown logic.
+ */
+ @VisibleForTesting
Review Comment:
You have marked this as `@VisibleForTesting`, but it is not called from any tests, can you make it private?
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: issues-unsubscribe@flink.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [flink] dzikosc commented on a diff in pull request #20133: Kinesis Consumer stop-with-savepoint support including EFO
Posted by GitBox <gi...@apache.org>.
dzikosc commented on code in PR #20133:
URL: https://github.com/apache/flink/pull/20133#discussion_r913080950
##########
flink-connectors/flink-connector-kinesis/src/main/java/org/apache/flink/streaming/connectors/kinesis/internals/ShardConsumer.java:
##########
@@ -149,7 +149,9 @@ public void run() {
// we can close this consumer thread once we've reached the end of the
// subscribed shard
break;
- } else if (result == CANCELLED) {
+ } else if (result == CANCELLED && isRunning()) {
+ // cancellation might be signalled as part of graceful shutdown during
+ // stop-with-savepoint, we should ignore it when the consumer is not running
Review Comment:
Removed the notion of stop-with-savepoint
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: issues-unsubscribe@flink.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [flink] dannycranmer commented on pull request #20133: [FLINK-23528][connectors/kinesis] Kinesis Consumer stop-with-savepoint support including EFO
Posted by GitBox <gi...@apache.org>.
dannycranmer commented on PR #20133:
URL: https://github.com/apache/flink/pull/20133#issuecomment-1190502240
@flinkbot run azure
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: issues-unsubscribe@flink.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [flink] dannycranmer commented on pull request #20133: [FLINK-23528][connectors/kinesis] Kinesis Consumer stop-with-savepoint support including EFO
Posted by GitBox <gi...@apache.org>.
dannycranmer commented on PR #20133:
URL: https://github.com/apache/flink/pull/20133#issuecomment-1190799792
@flinkbot run azure
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: issues-unsubscribe@flink.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [flink] dzikosc commented on a diff in pull request #20133: Kinesis Consumer stop-with-savepoint support including EFO
Posted by GitBox <gi...@apache.org>.
dzikosc commented on code in PR #20133:
URL: https://github.com/apache/flink/pull/20133#discussion_r913028922
##########
flink-connectors/flink-connector-kinesis/src/test/java/org/apache/flink/streaming/connectors/kinesis/FlinkKinesisITCase.java:
##########
@@ -69,104 +81,174 @@ public class FlinkKinesisITCase extends TestLogger {
@Rule public TemporaryFolder temp = new TemporaryFolder();
+ @Rule public SharedObjects sharedObjects = SharedObjects.create();
+
private static final SimpleStringSchema STRING_SCHEMA = new SimpleStringSchema();
private KinesisPubsubClient client;
@Before
- public void setupClient() {
- client = new KinesisPubsubClient(kinesalite.getContainerProperties());
+ public void setupClient() throws Exception {
+ client = new KinesisPubsubClient(getContainerProperties());
+ stream = TestNameProvider.getCurrentTestName().replaceAll("\\W", "");
+ client.createTopic(stream, 1, new Properties());
+ }
+
+ @Test
+ public void testStopWithSavepoint() throws Exception {
+ testStopWithSavepoint(false, false);
+ }
+
+ @Test
+ public void testStopWithSavepointWithDrain() throws Exception {
+ testStopWithSavepoint(true, false);
+ }
+
+ @Test
+ @Ignore("Kinesalite does not support EFO")
+ public void testStopWithSavepointWithEfo() throws Exception {
+ testStopWithSavepoint(false, true);
+ }
+
+ @Test
+ @Ignore("Kinesalite does not support EFO")
+ public void testStopWithSavepointWithDrainAndEfo() throws Exception {
+ testStopWithSavepoint(true, true);
}
/**
* Tests that pending elements do not cause a deadlock during stop with savepoint (FLINK-17170).
*
* <ol>
- * <li>The test setups up a stream with 100 records and creates a Flink job that reads them
+ * <li>The test setups up a stream with 1000 records and creates a Flink job that reads them
* with very slowly (using up a large chunk of time of the mailbox).
* <li>After ensuring that consumption has started, the job is stopped in a parallel thread.
* <li>Without the fix of FLINK-17170, the job now has a high chance to deadlock during
* cancel.
* <li>With the fix, the job proceeds and we can lift the backpressure.
* </ol>
*/
- @Test
- public void testStopWithSavepoint() throws Exception {
- client.createTopic(TEST_STREAM, 1, new Properties());
-
+ private void testStopWithSavepoint(boolean drain, boolean efo) throws Exception {
// add elements to the test stream
int numElements = 1000;
client.sendMessage(
- TEST_STREAM,
+ stream,
IntStream.range(0, numElements).mapToObj(String::valueOf).toArray(String[]::new));
StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();
env.setParallelism(1);
+ env.enableCheckpointing(100L);
- Properties config = kinesalite.getContainerProperties();
- config.setProperty(STREAM_INITIAL_POSITION, InitialPosition.TRIM_HORIZON.name());
- FlinkKinesisConsumer<String> consumer =
- new FlinkKinesisConsumer<>(TEST_STREAM, STRING_SCHEMA, config);
+ SharedReference<CountDownLatch> savepointTrigger = sharedObjects.add(new CountDownLatch(1));
+ DataStream<String> outputStream =
+ env.addSource(createKinesisConsumer(efo)).map(new WaitingMapper(savepointTrigger));
- DataStream<String> stream = env.addSource(consumer).map(new WaitingMapper());
// call stop with savepoint in another thread
ForkJoinTask<Object> stopTask =
ForkJoinPool.commonPool()
.submit(
() -> {
- WaitingMapper.firstElement.await();
- stopWithSavepoint();
- WaitingMapper.stopped = true;
+ savepointTrigger.get().await();
+ stopWithSavepoint(drain);
return null;
});
try {
- List<String> result = stream.executeAndCollect(10000);
+ List<String> result = outputStream.executeAndCollect(10000);
// stop with savepoint will most likely only return a small subset of the elements
// validate that the prefix is as expected
- assertThat(result, hasSize(lessThan(numElements)));
- assertThat(
- result,
- equalTo(
- IntStream.range(0, numElements)
- .mapToObj(String::valueOf)
- .collect(Collectors.toList())
- .subList(0, result.size())));
+ if (drain) {
+ assertThat(
+ result,
+ contains(
+ IntStream.range(0, numElements)
+ .mapToObj(String::valueOf)
+ .toArray()));
+ } else {
+ // stop with savepoint will most likely only return a small subset of the elements
+ // validate that the prefix is as expected
+ assertThat(
+ result,
+ contains(
+ IntStream.range(0, result.size())
+ .mapToObj(String::valueOf)
+ .toArray()));
+ }
} finally {
- stopTask.cancel(true);
+ stopTask.get();
+ }
+ }
+
+ private FlinkKinesisConsumer<String> createKinesisConsumer(boolean efo) {
+ Properties config = getContainerProperties();
+ config.setProperty(STREAM_INITIAL_POSITION, InitialPosition.TRIM_HORIZON.name());
+ if (efo) {
+ config.putIfAbsent(RECORD_PUBLISHER_TYPE, "EFO");
+ config.putIfAbsent(EFO_CONSUMER_NAME, "efo-flink-app");
}
+ return new FlinkKinesisConsumer<>(stream, STRING_SCHEMA, config);
+ }
+
+ private Properties getContainerProperties() {
+ return kinesalite.getContainerProperties();
}
- private String stopWithSavepoint() throws Exception {
+ private String stopWithSavepoint(boolean drain) throws Exception {
JobStatusMessage job =
MINI_CLUSTER.getClusterClient().listJobs().get().stream().findFirst().get();
return MINI_CLUSTER
.getClusterClient()
.stopWithSavepoint(
job.getJobId(),
- true,
+ drain,
temp.getRoot().getAbsolutePath(),
SavepointFormatType.CANONICAL)
.get();
}
- private static class WaitingMapper implements MapFunction<String, String> {
- static CountDownLatch firstElement;
- static volatile boolean stopped;
+ private static class WaitingMapper
+ implements MapFunction<String, String>, CheckpointedFunction {
+ private final SharedReference<CountDownLatch> savepointTrigger;
+ private volatile boolean savepointTriggered;
+ // keep track on when the last checkpoint occurred
+ private transient Deadline checkpointDeadline;
+ private final AtomicInteger numElements = new AtomicInteger();
+
+ WaitingMapper(SharedReference<CountDownLatch> savepointTrigger) {
+ this.savepointTrigger = savepointTrigger;
+ checkpointDeadline = Deadline.fromNow(Duration.ofDays(1));
Review Comment:
Left a comment explaining some of the context.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: issues-unsubscribe@flink.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [flink] dannycranmer commented on pull request #20133: [FLINK-23528][connectors/kinesis] Kinesis Consumer stop-with-savepoint support including EFO
Posted by GitBox <gi...@apache.org>.
dannycranmer commented on PR #20133:
URL: https://github.com/apache/flink/pull/20133#issuecomment-1193774460
Thanks @dzikosc, LGTM, merging.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: issues-unsubscribe@flink.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [flink] dzikosc commented on pull request #20133: Kinesis Consumer stop-with-savepoint support including EFO
Posted by GitBox <gi...@apache.org>.
dzikosc commented on PR #20133:
URL: https://github.com/apache/flink/pull/20133#issuecomment-1172434446
This is a quick first PR to let the tests run in CI pipe
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: issues-unsubscribe@flink.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [flink] dannycranmer commented on a diff in pull request #20133: Kinesis Consumer stop-with-savepoint support including EFO
Posted by GitBox <gi...@apache.org>.
dannycranmer commented on code in PR #20133:
URL: https://github.com/apache/flink/pull/20133#discussion_r912129547
##########
flink-connectors/flink-connector-kinesis/src/main/java/org/apache/flink/streaming/connectors/kinesis/internals/ShardConsumer.java:
##########
@@ -149,7 +149,9 @@ public void run() {
// we can close this consumer thread once we've reached the end of the
// subscribed shard
break;
- } else if (result == CANCELLED) {
+ } else if (result == CANCELLED && isRunning()) {
+ // cancellation might be signalled as part of graceful shutdown during
+ // stop-with-savepoint, we should ignore it when the consumer is not running
final String errorMessage =
"Shard consumer cancelled: " + subscribedShard.getShard().getShardId();
Review Comment:
Can you please add a unit test for this change
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: issues-unsubscribe@flink.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [flink] dzikosc commented on pull request #20133: Kinesis Consumer stop-with-savepoint support including EFO
Posted by GitBox <gi...@apache.org>.
dzikosc commented on PR #20133:
URL: https://github.com/apache/flink/pull/20133#issuecomment-1172555651
@flinkbot run azure
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: issues-unsubscribe@flink.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [flink] dzikosc commented on a diff in pull request #20133: Kinesis Consumer stop-with-savepoint support including EFO
Posted by GitBox <gi...@apache.org>.
dzikosc commented on code in PR #20133:
URL: https://github.com/apache/flink/pull/20133#discussion_r913032001
##########
flink-streaming-java/src/main/java/org/apache/flink/streaming/api/operators/collect/CollectSinkOperator.java:
##########
@@ -51,9 +51,9 @@ public void handleOperatorEvent(OperatorEvent evt) {
}
@Override
- public void finish() throws Exception {
+ public void close() throws Exception {
sinkFunction.accumulateFinalResults();
- super.finish();
+ super.close();
}
Review Comment:
The change has been originally contributed by @AHeise. I haven't evaluated this in depth, however the original commit message suggests:
> DataStream#executeAndCollect expects the CollectSinkOperator to register the accumulator at the end of the application or fails with some exception.
However, a stop-with-savepoint without drain would not trigger CollectSinkOperator#finish and thus skip the registration.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: issues-unsubscribe@flink.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [flink] dannycranmer commented on pull request #20133: Kinesis Consumer stop-with-savepoint support including EFO
Posted by GitBox <gi...@apache.org>.
dannycranmer commented on PR #20133:
URL: https://github.com/apache/flink/pull/20133#issuecomment-1190399646
LGTM, will merge once CI passes
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: issues-unsubscribe@flink.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [flink] flinkbot commented on pull request #20133: Kinesis Consumer stop-with-savepoint support including EFO
Posted by GitBox <gi...@apache.org>.
flinkbot commented on PR #20133:
URL: https://github.com/apache/flink/pull/20133#issuecomment-1172433681
<!--
Meta data
{
"version" : 1,
"metaDataEntries" : [ {
"hash" : "381882d42197d8ef869812518bbee3c2df5c178c",
"status" : "UNKNOWN",
"url" : "TBD",
"triggerID" : "381882d42197d8ef869812518bbee3c2df5c178c",
"triggerType" : "PUSH"
} ]
}-->
## CI report:
* 381882d42197d8ef869812518bbee3c2df5c178c UNKNOWN
<details>
<summary>Bot commands</summary>
The @flinkbot bot supports the following commands:
- `@flinkbot run azure` re-run the last Azure build
</details>
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: issues-unsubscribe@flink.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [flink] CrynetLogistics commented on pull request #20133: Kinesis Consumer stop-with-savepoint support including EFO
Posted by GitBox <gi...@apache.org>.
CrynetLogistics commented on PR #20133:
URL: https://github.com/apache/flink/pull/20133#issuecomment-1172445207
@flinkbot run azure
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: issues-unsubscribe@flink.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [flink] dannycranmer commented on a diff in pull request #20133: Kinesis Consumer stop-with-savepoint support including EFO
Posted by GitBox <gi...@apache.org>.
dannycranmer commented on code in PR #20133:
URL: https://github.com/apache/flink/pull/20133#discussion_r912126766
##########
flink-streaming-java/src/main/java/org/apache/flink/streaming/api/operators/collect/CollectSinkOperator.java:
##########
@@ -51,9 +51,9 @@ public void handleOperatorEvent(OperatorEvent evt) {
}
@Override
- public void finish() throws Exception {
+ public void close() throws Exception {
sinkFunction.accumulateFinalResults();
- super.finish();
+ super.close();
}
Review Comment:
Why did we need to change this? Does this impact any other connectors?
##########
flink-connectors/flink-connector-kinesis/src/main/java/org/apache/flink/streaming/connectors/kinesis/internals/ShardConsumer.java:
##########
@@ -149,7 +149,9 @@ public void run() {
// we can close this consumer thread once we've reached the end of the
// subscribed shard
break;
- } else if (result == CANCELLED) {
+ } else if (result == CANCELLED && isRunning()) {
+ // cancellation might be signalled as part of graceful shutdown during
+ // stop-with-savepoint, we should ignore it when the consumer is not running
Review Comment:
nit: Since we have changed the `shutdownFetcher` method, I think `running == false` is going to be the new normal shutdown shutdown procedure for EFO
##########
flink-connectors/flink-connector-kinesis/src/test/java/org/apache/flink/streaming/connectors/kinesis/FlinkKinesisITCase.java:
##########
@@ -69,104 +81,174 @@ public class FlinkKinesisITCase extends TestLogger {
@Rule public TemporaryFolder temp = new TemporaryFolder();
+ @Rule public SharedObjects sharedObjects = SharedObjects.create();
+
private static final SimpleStringSchema STRING_SCHEMA = new SimpleStringSchema();
private KinesisPubsubClient client;
@Before
- public void setupClient() {
- client = new KinesisPubsubClient(kinesalite.getContainerProperties());
+ public void setupClient() throws Exception {
+ client = new KinesisPubsubClient(getContainerProperties());
+ stream = TestNameProvider.getCurrentTestName().replaceAll("\\W", "");
+ client.createTopic(stream, 1, new Properties());
+ }
+
+ @Test
+ public void testStopWithSavepoint() throws Exception {
+ testStopWithSavepoint(false, false);
+ }
+
+ @Test
+ public void testStopWithSavepointWithDrain() throws Exception {
+ testStopWithSavepoint(true, false);
+ }
+
+ @Test
+ @Ignore("Kinesalite does not support EFO")
+ public void testStopWithSavepointWithEfo() throws Exception {
+ testStopWithSavepoint(false, true);
+ }
+
+ @Test
+ @Ignore("Kinesalite does not support EFO")
+ public void testStopWithSavepointWithDrainAndEfo() throws Exception {
+ testStopWithSavepoint(true, true);
}
/**
* Tests that pending elements do not cause a deadlock during stop with savepoint (FLINK-17170).
*
* <ol>
- * <li>The test setups up a stream with 100 records and creates a Flink job that reads them
+ * <li>The test setups up a stream with 1000 records and creates a Flink job that reads them
* with very slowly (using up a large chunk of time of the mailbox).
* <li>After ensuring that consumption has started, the job is stopped in a parallel thread.
* <li>Without the fix of FLINK-17170, the job now has a high chance to deadlock during
* cancel.
* <li>With the fix, the job proceeds and we can lift the backpressure.
* </ol>
*/
- @Test
- public void testStopWithSavepoint() throws Exception {
- client.createTopic(TEST_STREAM, 1, new Properties());
-
+ private void testStopWithSavepoint(boolean drain, boolean efo) throws Exception {
// add elements to the test stream
int numElements = 1000;
client.sendMessage(
- TEST_STREAM,
+ stream,
IntStream.range(0, numElements).mapToObj(String::valueOf).toArray(String[]::new));
StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();
env.setParallelism(1);
+ env.enableCheckpointing(100L);
- Properties config = kinesalite.getContainerProperties();
- config.setProperty(STREAM_INITIAL_POSITION, InitialPosition.TRIM_HORIZON.name());
- FlinkKinesisConsumer<String> consumer =
- new FlinkKinesisConsumer<>(TEST_STREAM, STRING_SCHEMA, config);
+ SharedReference<CountDownLatch> savepointTrigger = sharedObjects.add(new CountDownLatch(1));
+ DataStream<String> outputStream =
+ env.addSource(createKinesisConsumer(efo)).map(new WaitingMapper(savepointTrigger));
- DataStream<String> stream = env.addSource(consumer).map(new WaitingMapper());
// call stop with savepoint in another thread
ForkJoinTask<Object> stopTask =
ForkJoinPool.commonPool()
.submit(
() -> {
- WaitingMapper.firstElement.await();
- stopWithSavepoint();
- WaitingMapper.stopped = true;
+ savepointTrigger.get().await();
+ stopWithSavepoint(drain);
return null;
});
try {
- List<String> result = stream.executeAndCollect(10000);
+ List<String> result = outputStream.executeAndCollect(10000);
// stop with savepoint will most likely only return a small subset of the elements
// validate that the prefix is as expected
- assertThat(result, hasSize(lessThan(numElements)));
- assertThat(
- result,
- equalTo(
- IntStream.range(0, numElements)
- .mapToObj(String::valueOf)
- .collect(Collectors.toList())
- .subList(0, result.size())));
+ if (drain) {
+ assertThat(
+ result,
+ contains(
+ IntStream.range(0, numElements)
+ .mapToObj(String::valueOf)
+ .toArray()));
+ } else {
+ // stop with savepoint will most likely only return a small subset of the elements
+ // validate that the prefix is as expected
+ assertThat(
+ result,
+ contains(
+ IntStream.range(0, result.size())
+ .mapToObj(String::valueOf)
+ .toArray()));
+ }
} finally {
- stopTask.cancel(true);
+ stopTask.get();
+ }
+ }
+
+ private FlinkKinesisConsumer<String> createKinesisConsumer(boolean efo) {
+ Properties config = getContainerProperties();
+ config.setProperty(STREAM_INITIAL_POSITION, InitialPosition.TRIM_HORIZON.name());
+ if (efo) {
+ config.putIfAbsent(RECORD_PUBLISHER_TYPE, "EFO");
+ config.putIfAbsent(EFO_CONSUMER_NAME, "efo-flink-app");
}
+ return new FlinkKinesisConsumer<>(stream, STRING_SCHEMA, config);
+ }
+
+ private Properties getContainerProperties() {
+ return kinesalite.getContainerProperties();
}
- private String stopWithSavepoint() throws Exception {
+ private String stopWithSavepoint(boolean drain) throws Exception {
JobStatusMessage job =
MINI_CLUSTER.getClusterClient().listJobs().get().stream().findFirst().get();
return MINI_CLUSTER
.getClusterClient()
.stopWithSavepoint(
job.getJobId(),
- true,
+ drain,
temp.getRoot().getAbsolutePath(),
SavepointFormatType.CANONICAL)
.get();
}
- private static class WaitingMapper implements MapFunction<String, String> {
- static CountDownLatch firstElement;
- static volatile boolean stopped;
+ private static class WaitingMapper
+ implements MapFunction<String, String>, CheckpointedFunction {
+ private final SharedReference<CountDownLatch> savepointTrigger;
+ private volatile boolean savepointTriggered;
+ // keep track on when the last checkpoint occurred
+ private transient Deadline checkpointDeadline;
+ private final AtomicInteger numElements = new AtomicInteger();
+
+ WaitingMapper(SharedReference<CountDownLatch> savepointTrigger) {
+ this.savepointTrigger = savepointTrigger;
+ checkpointDeadline = Deadline.fromNow(Duration.ofDays(1));
+ }
- WaitingMapper() {
- firstElement = new CountDownLatch(1);
- stopped = false;
+ private void readObject(ObjectInputStream stream)
+ throws ClassNotFoundException, IOException {
+ stream.defaultReadObject();
+ checkpointDeadline = Deadline.fromNow(Duration.ofDays(1));
}
@Override
public String map(String value) throws Exception {
- if (firstElement.getCount() > 0) {
- firstElement.countDown();
- }
- if (!stopped) {
- Thread.sleep(100);
+ numElements.incrementAndGet();
+ if (!savepointTriggered) {
+ try {
+ Thread.sleep(100);
+ } catch (InterruptedException e) {
Review Comment:
Is swallowing the exception the right thing todo? Can you add a comment to explain this?
##########
flink-connectors/flink-connector-kinesis/src/test/java/org/apache/flink/streaming/connectors/kinesis/FlinkKinesisITCase.java:
##########
@@ -69,104 +81,174 @@ public class FlinkKinesisITCase extends TestLogger {
@Rule public TemporaryFolder temp = new TemporaryFolder();
+ @Rule public SharedObjects sharedObjects = SharedObjects.create();
+
private static final SimpleStringSchema STRING_SCHEMA = new SimpleStringSchema();
private KinesisPubsubClient client;
@Before
- public void setupClient() {
- client = new KinesisPubsubClient(kinesalite.getContainerProperties());
+ public void setupClient() throws Exception {
+ client = new KinesisPubsubClient(getContainerProperties());
+ stream = TestNameProvider.getCurrentTestName().replaceAll("\\W", "");
+ client.createTopic(stream, 1, new Properties());
+ }
+
+ @Test
+ public void testStopWithSavepoint() throws Exception {
+ testStopWithSavepoint(false, false);
+ }
+
+ @Test
+ public void testStopWithSavepointWithDrain() throws Exception {
+ testStopWithSavepoint(true, false);
+ }
+
+ @Test
+ @Ignore("Kinesalite does not support EFO")
+ public void testStopWithSavepointWithEfo() throws Exception {
+ testStopWithSavepoint(false, true);
+ }
+
+ @Test
+ @Ignore("Kinesalite does not support EFO")
+ public void testStopWithSavepointWithDrainAndEfo() throws Exception {
+ testStopWithSavepoint(true, true);
}
/**
* Tests that pending elements do not cause a deadlock during stop with savepoint (FLINK-17170).
*
* <ol>
- * <li>The test setups up a stream with 100 records and creates a Flink job that reads them
+ * <li>The test setups up a stream with 1000 records and creates a Flink job that reads them
* with very slowly (using up a large chunk of time of the mailbox).
* <li>After ensuring that consumption has started, the job is stopped in a parallel thread.
* <li>Without the fix of FLINK-17170, the job now has a high chance to deadlock during
* cancel.
* <li>With the fix, the job proceeds and we can lift the backpressure.
* </ol>
*/
- @Test
- public void testStopWithSavepoint() throws Exception {
- client.createTopic(TEST_STREAM, 1, new Properties());
-
+ private void testStopWithSavepoint(boolean drain, boolean efo) throws Exception {
// add elements to the test stream
int numElements = 1000;
client.sendMessage(
- TEST_STREAM,
+ stream,
IntStream.range(0, numElements).mapToObj(String::valueOf).toArray(String[]::new));
StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();
env.setParallelism(1);
+ env.enableCheckpointing(100L);
- Properties config = kinesalite.getContainerProperties();
- config.setProperty(STREAM_INITIAL_POSITION, InitialPosition.TRIM_HORIZON.name());
- FlinkKinesisConsumer<String> consumer =
- new FlinkKinesisConsumer<>(TEST_STREAM, STRING_SCHEMA, config);
+ SharedReference<CountDownLatch> savepointTrigger = sharedObjects.add(new CountDownLatch(1));
+ DataStream<String> outputStream =
+ env.addSource(createKinesisConsumer(efo)).map(new WaitingMapper(savepointTrigger));
- DataStream<String> stream = env.addSource(consumer).map(new WaitingMapper());
// call stop with savepoint in another thread
ForkJoinTask<Object> stopTask =
ForkJoinPool.commonPool()
.submit(
() -> {
- WaitingMapper.firstElement.await();
- stopWithSavepoint();
- WaitingMapper.stopped = true;
+ savepointTrigger.get().await();
+ stopWithSavepoint(drain);
return null;
});
try {
- List<String> result = stream.executeAndCollect(10000);
+ List<String> result = outputStream.executeAndCollect(10000);
// stop with savepoint will most likely only return a small subset of the elements
// validate that the prefix is as expected
- assertThat(result, hasSize(lessThan(numElements)));
- assertThat(
- result,
- equalTo(
- IntStream.range(0, numElements)
- .mapToObj(String::valueOf)
- .collect(Collectors.toList())
- .subList(0, result.size())));
+ if (drain) {
+ assertThat(
+ result,
+ contains(
+ IntStream.range(0, numElements)
+ .mapToObj(String::valueOf)
+ .toArray()));
+ } else {
+ // stop with savepoint will most likely only return a small subset of the elements
+ // validate that the prefix is as expected
+ assertThat(
+ result,
+ contains(
+ IntStream.range(0, result.size())
+ .mapToObj(String::valueOf)
+ .toArray()));
+ }
} finally {
- stopTask.cancel(true);
+ stopTask.get();
+ }
+ }
+
+ private FlinkKinesisConsumer<String> createKinesisConsumer(boolean efo) {
+ Properties config = getContainerProperties();
+ config.setProperty(STREAM_INITIAL_POSITION, InitialPosition.TRIM_HORIZON.name());
+ if (efo) {
+ config.putIfAbsent(RECORD_PUBLISHER_TYPE, "EFO");
+ config.putIfAbsent(EFO_CONSUMER_NAME, "efo-flink-app");
}
+ return new FlinkKinesisConsumer<>(stream, STRING_SCHEMA, config);
+ }
+
+ private Properties getContainerProperties() {
+ return kinesalite.getContainerProperties();
}
- private String stopWithSavepoint() throws Exception {
+ private String stopWithSavepoint(boolean drain) throws Exception {
JobStatusMessage job =
MINI_CLUSTER.getClusterClient().listJobs().get().stream().findFirst().get();
return MINI_CLUSTER
.getClusterClient()
.stopWithSavepoint(
job.getJobId(),
- true,
+ drain,
temp.getRoot().getAbsolutePath(),
SavepointFormatType.CANONICAL)
.get();
}
- private static class WaitingMapper implements MapFunction<String, String> {
- static CountDownLatch firstElement;
- static volatile boolean stopped;
+ private static class WaitingMapper
+ implements MapFunction<String, String>, CheckpointedFunction {
+ private final SharedReference<CountDownLatch> savepointTrigger;
+ private volatile boolean savepointTriggered;
+ // keep track on when the last checkpoint occurred
+ private transient Deadline checkpointDeadline;
+ private final AtomicInteger numElements = new AtomicInteger();
+
+ WaitingMapper(SharedReference<CountDownLatch> savepointTrigger) {
+ this.savepointTrigger = savepointTrigger;
+ checkpointDeadline = Deadline.fromNow(Duration.ofDays(1));
Review Comment:
Why is this deadline so high?
##########
flink-connectors/flink-connector-kinesis/src/main/java/org/apache/flink/streaming/connectors/kinesis/internals/publisher/fanout/FanOutRecordPublisher.java:
##########
@@ -154,7 +155,8 @@ private RecordPublisherRunResult runWithBackoff(
fanOutShardSubscriber.subscribeToShardAndConsumeRecords(
toSdkV2StartingPosition(nextStartingPosition), eventConsumer);
attempt = 0;
- } catch (FanOutSubscriberInterruptedException ex) {
+ } catch (FanOutSubscriberInterruptedException | RejectedExecutionException ex) {
+ // RejectedExecutionException occurs during soft shutdown, e.g. stop with savepoint
Review Comment:
Did we also test stop without savepoint?
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: issues-unsubscribe@flink.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [flink] dannycranmer merged pull request #20133: [FLINK-23528][connectors/kinesis] Kinesis Consumer stop-with-savepoint support including EFO
Posted by GitBox <gi...@apache.org>.
dannycranmer merged PR #20133:
URL: https://github.com/apache/flink/pull/20133
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: issues-unsubscribe@flink.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org