You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@flink.apache.org by tr...@apache.org on 2021/07/02 15:46:49 UTC
[flink] branch release-1.13 updated: [hotfix][release-notes] Update
release notes for 1.13 with accumulators semantics change in
MiniClusterJobClient
This is an automated email from the ASF dual-hosted git repository.
trohrmann pushed a commit to branch release-1.13
in repository https://gitbox.apache.org/repos/asf/flink.git
The following commit(s) were added to refs/heads/release-1.13 by this push:
new 5c4be8a [hotfix][release-notes] Update release notes for 1.13 with accumulators semantics change in MiniClusterJobClient
5c4be8a is described below
commit 5c4be8a89643422d2335243a1a58edfa45ea4721
Author: Etienne Chauchot <ec...@apache.org>
AuthorDate: Wed Jun 23 12:24:14 2021 +0200
[hotfix][release-notes] Update release notes for 1.13 with accumulators semantics change in MiniClusterJobClient
FLINK-18685 changed the semantics of the MiniClusterJobClient. This commit updates the 1.13 release notes
accordingly.
This closes #16256.
---
docs/content.zh/release-notes/flink-1.11.md | 2 +-
docs/content.zh/release-notes/flink-1.13.md | 12 ++++++++++++
docs/content/release-notes/flink-1.13.md | 12 ++++++++++++
3 files changed, 25 insertions(+), 1 deletion(-)
diff --git a/docs/content.zh/release-notes/flink-1.11.md b/docs/content.zh/release-notes/flink-1.11.md
index fcc325c..2807dd4 100644
--- a/docs/content.zh/release-notes/flink-1.11.md
+++ b/docs/content.zh/release-notes/flink-1.11.md
@@ -384,4 +384,4 @@ Use the `CheckpointedFunction` interface instead, which gives more control over
We removed deprecated state access methods `RuntimeContext#getFoldingState()`, `OperatorStateStore#getSerializableListState()` and `OperatorStateStore#getOperatorState()`.
This means that some code that was compiled against Flink 1.10 will not work with a Flink 1.11 cluster.
-An example of this is our Kafka connector which internally used `OperatorStateStore.getSerializableListState`.
\ No newline at end of file
+An example of this is our Kafka connector which internally used `OperatorStateStore.getSerializableListState`.
diff --git a/docs/content.zh/release-notes/flink-1.13.md b/docs/content.zh/release-notes/flink-1.13.md
index 6081f9b..13def89 100644
--- a/docs/content.zh/release-notes/flink-1.13.md
+++ b/docs/content.zh/release-notes/flink-1.13.md
@@ -191,6 +191,18 @@ This feature is not covering all locations in Flink where user code is executed.
infrastructure for such an interception. We are tracking this improvement in
[FLINK-21307](https://issues.apache.org/jira/browse/FLINK-21307).
+#### MiniClusterJobClient#getAccumulators was infinitely blocking in local environment for a streaming job
+##### [FLINK-18685](https://issues.apache.org/jira/browse/FLINK-18685)
+
+The semantics for accumulators have now changed in `MiniClusterJobClient` to fix this bug and comply with other JobClient implementations:
+Previously `MiniClusterJobClient` assumed that `getAccumulator()` was called on a bounded pipeline and that the user wanted to acquire the final
+accumulator values after the job is finished.
+But now it returns the current value of accumulators immediately to be compatible with unbounded pipelines.
+
+If it is run on a bounded pipeline, then to get the final accumulator values after the job is finished, one needs to call
+
+`getJobExecutionResult().thenApply(JobExecutionResult::getAllAccumulatorResults)`
+
### Docker
#### Consider removing automatic configuration fo number of slots from docker
diff --git a/docs/content/release-notes/flink-1.13.md b/docs/content/release-notes/flink-1.13.md
index 6081f9b..13def89 100644
--- a/docs/content/release-notes/flink-1.13.md
+++ b/docs/content/release-notes/flink-1.13.md
@@ -191,6 +191,18 @@ This feature is not covering all locations in Flink where user code is executed.
infrastructure for such an interception. We are tracking this improvement in
[FLINK-21307](https://issues.apache.org/jira/browse/FLINK-21307).
+#### MiniClusterJobClient#getAccumulators was infinitely blocking in local environment for a streaming job
+##### [FLINK-18685](https://issues.apache.org/jira/browse/FLINK-18685)
+
+The semantics for accumulators have now changed in `MiniClusterJobClient` to fix this bug and comply with other JobClient implementations:
+Previously `MiniClusterJobClient` assumed that `getAccumulator()` was called on a bounded pipeline and that the user wanted to acquire the final
+accumulator values after the job is finished.
+But now it returns the current value of accumulators immediately to be compatible with unbounded pipelines.
+
+If it is run on a bounded pipeline, then to get the final accumulator values after the job is finished, one needs to call
+
+`getJobExecutionResult().thenApply(JobExecutionResult::getAllAccumulatorResults)`
+
### Docker
#### Consider removing automatic configuration fo number of slots from docker