lirui-apache commented on a change in pull request #14310:
URL: https://github.com/apache/flink/pull/14310#discussion_r537199758
##########
File path:
flink-connectors/flink-connector-hive/src/main/java/org/apache/flink/connectors/hive/HiveLookupTableSource.java
##########
@@ -114,16 +117,16 @@ private void validateLookupConfigurations() {
Duration monitorInterval =
configuration.get(STREAMING_SOURCE_MONITOR_INTERVAL) == null
? DEFAULT_LOOKUP_MONITOR_INTERVAL
:
configuration.get(STREAMING_SOURCE_MONITOR_INTERVAL);
- Preconditions.checkArgument(
- monitorInterval.toMillis() >=
DEFAULT_LOOKUP_MONITOR_INTERVAL.toMillis(),
- String.format(
- "Currently the value of
'%s' is required bigger or equal to default value '%s' " +
- "when
set '%s' to 'latest', but actual is '%s'",
-
STREAMING_SOURCE_MONITOR_INTERVAL.key(),
-
DEFAULT_LOOKUP_MONITOR_INTERVAL.toMillis(),
-
STREAMING_SOURCE_PARTITION_INCLUDE.key(),
-
monitorInterval.toMillis())
- );
+
+ if (monitorInterval.toMillis() <
DEFAULT_LOOKUP_MONITOR_INTERVAL.toMillis()) {
+ LOG.warn(String.format(
+ "Currently the recommended value of
'%s' is bigger than default value '%s' " +
Review comment:
I agree `1h` may be more popular than `30min` for partitioning. But just
because more users use `1h` doesn't mean those using `30min` deserve a scary
warning.
When writing stream data into hive, we usually tell users that the minimum
value supported for `sink.partition-commit.delay` is around `10min`. So I
suppose the warning threshold here shouldn't be bigger than that. Because if
users can commit new partitions every 10min, it's reasonable to assume they
also want to monitor at roughly the same interval. WDYT?
----------------------------------------------------------------
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.
For queries about this service, please contact Infrastructure at:
[email protected]
