FrankChen021 commented on a change in pull request #10551:
URL: https://github.com/apache/druid/pull/10551#discussion_r518654741
##########
File path:
extensions-core/kafka-indexing-service/src/main/java/org/apache/druid/indexing/kafka/KafkaConsumerConfigs.java
##########
@@ -38,7 +38,6 @@
props.put("group.id", StringUtils.format("kafka-supervisor-%s",
IdUtils.getRandomId()));
props.put("auto.offset.reset", "none");
props.put("enable.auto.commit", "false");
- props.put("isolation.level", "read_committed");
Review comment:
I think to remove `isolation.level` here and leaves the configuration to
users changes the current behavior.
Currently users care nothing about this kafka configuration when using a
higher version of kafka such as 0.11. By removing it, they need to set this
property in supervisor spec, or the default value, which is
`read_uncommitted`, will be applied, which may be not what they expect.
If this is the root cause that limits the Druid to use Kafka lower than
0.11, I think maybe we can introduce another property, as the same way as
`pollTimeout` property does, then we can unset `isolation.level` property
according to the value of this new property. Only those who want to use Druid
with older kafka need to set this new property.
----------------------------------------------------------------
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]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
