pnowojski commented on code in PR #23425:
URL: https://github.com/apache/flink/pull/23425#discussion_r1365013893
##########
flink-core/src/main/java/org/apache/flink/configuration/CheckpointingOptions.java:
##########
@@ -109,6 +109,20 @@ public class CheckpointingOptions {
.defaultValue(1)
.withDescription("The maximum number of completed
checkpoints to retain.");
+ /**
+ * Option whether to clean individual checkpoint's operatorstates in
parallel. If enabled,
+ * operator states are discarded in parallel using the ExecutorService
passed to the cleaner.
+ * This speeds up checkpoints cleaning, but adds load to the IO.
+ */
+ @Documentation.Section(Documentation.Sections.COMMON_STATE_BACKENDS)
+ public static final ConfigOption<Boolean> CLEANER_PARALLEL_MODE =
+ ConfigOptions.key("state.checkpoint.cleaner.parallel-mode")
+ .booleanType()
+ .defaultValue(false)
Review Comment:
I would keep more or else the old code path, for example if there is a bug
or some unexpected behaviour, so that users could disable this feature.
Otherwise I think it should be enabled by default and if nobody complains, in
next or the following release we could drop the flag to force sequential clean
up (so we would only have parallel clean up).
--
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: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
