vvcephei commented on a change in pull request #8541:
URL: https://github.com/apache/kafka/pull/8541#discussion_r414195680
##########
File path:
streams/src/test/java/org/apache/kafka/streams/integration/EosIntegrationTest.java
##########
@@ -515,84 +520,114 @@ public void
shouldNotViolateEosIfOneTaskGetsFencedUsingIsolatedAppInstances() th
// the app is supposed to copy all 60 records into the output topic
// the app commits after each 10 records per partition, and thus will
have 2*5 uncommitted writes
//
- // a GC pause gets inject after 20 committed and 30 uncommitted
records got received
- // -> the GC pause only affects one thread and should trigger a
rebalance
+ // a stall gets injected after 20 committed and 30 uncommitted records
got received
+ // -> the stall only affects one thread and should trigger a rebalance
// after rebalancing, we should read 40 committed records (even if 50
record got written)
//
// afterwards, the "stalling" thread resumes, and another rebalance
should get triggered
// we write the remaining 20 records and verify to read 60 result
records
try (
- final KafkaStreams streams1 = getKafkaStreams(false, "appDir1", 1,
eosConfig);
- final KafkaStreams streams2 = getKafkaStreams(false, "appDir2", 1,
eosConfig)
+ final KafkaStreams streams1 = getKafkaStreams("streams1", false,
"appDir1", 1, eosConfig);
+ final KafkaStreams streams2 = getKafkaStreams("streams2", false,
"appDir2", 1, eosConfig)
Review comment:
I added an argument to the KafkaStreams builder to set the dummy host
name. Previously, it was always "dummy" even though we had two instances, which
resulted in the metadata map only containing one entry, even though there were
two nodes in the cluster. I'm not sure if this was a cause of flakiness (since
it seems it would be non-deterministic), but it's definitely not _right_.
----------------------------------------------------------------
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:
us...@infra.apache.org
