Alexey Serbin has posted comments on this change. ( 
http://gerrit.cloudera.org:8080/23246 )

Change subject: KUDU-3662 [5/n] Add metrics support to replication
......................................................................


Patch Set 4:

(5 comments)

Just a few questions and nits.

http://gerrit.cloudera.org:8080/#/c/23246/4//COMMIT_MSG
Commit Message:

http://gerrit.cloudera.org:8080/#/c/23246/4//COMMIT_MSG@11
PS4, Line 11: splits
nit: number of splits?


http://gerrit.cloudera.org:8080/#/c/23246/4//COMMIT_MSG@23
PS4, Line 23: unittests
nit: unit tests?


http://gerrit.cloudera.org:8080/#/c/23246/4/java/kudu-replication/src/main/java/org/apache/kudu/replication/wrappedsource/MetricWrappedKuduEnumerator.java
File 
java/kudu-replication/src/main/java/org/apache/kudu/replication/wrappedsource/MetricWrappedKuduEnumerator.java:

http://gerrit.cloudera.org:8080/#/c/23246/4/java/kudu-replication/src/main/java/org/apache/kudu/replication/wrappedsource/MetricWrappedKuduEnumerator.java@127
PS4, Line 127:   private long getLastEndTimestamp() {
             :     long hybridTime = 
ReflectionSecurityUtils.getLongFieldValue(lastEndTimestampField, delegate);
             :     long[] parsed = 
HybridTimeUtil.HTTimestampToPhysicalAndLogical(hybridTime);
             :     long epochMicros = parsed[0];
             :     return epochMicros / 1_000;
             :   }
I see that only the physical part of the timestamp is exposed via this metric, 
and I don't see any other metric in this changelist to expose the logical part. 
 What is the intuition/rationale behind not exposing it: in other words, how to 
understand that we aren't ever going to need it to tell on the progress of a 
replication job?


http://gerrit.cloudera.org:8080/#/c/23246/4/java/kudu-replication/src/test/java/org/apache/kudu/replication/TestReplicationMetrics.java
File 
java/kudu-replication/src/test/java/org/apache/kudu/replication/TestReplicationMetrics.java:

http://gerrit.cloudera.org:8080/#/c/23246/4/java/kudu-replication/src/test/java/org/apache/kudu/replication/TestReplicationMetrics.java@116
PS4, Line 116: flinkCluster.cancelAllJobs();
Quick question (nit) for here and below: is it OK not calling this if any of 
the assertions above triggers?


http://gerrit.cloudera.org:8080/#/c/23246/4/java/kudu-replication/src/test/java/org/apache/kudu/replication/TestReplicationMetrics.java@129
PS4, Line 129: 15000
Here the timeout is 15s, but below it's 10s.  Why is the difference and whether 
any flakiness expected with on busy machines when running this test?



--
To view, visit http://gerrit.cloudera.org:8080/23246
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-Project: kudu
Gerrit-Branch: master
Gerrit-MessageType: comment
Gerrit-Change-Id: Ibfbd34c707e7539ee88863399ae3061683f8bb3b
Gerrit-Change-Number: 23246
Gerrit-PatchSet: 4
Gerrit-Owner: Marton Greber <[email protected]>
Gerrit-Reviewer: Abhishek Chennaka <[email protected]>
Gerrit-Reviewer: Alexey Serbin <[email protected]>
Gerrit-Reviewer: Ashwani Raina <[email protected]>
Gerrit-Reviewer: Gabriella Lotz <[email protected]>
Gerrit-Reviewer: Kudu Jenkins (120)
Gerrit-Reviewer: Marton Greber <[email protected]>
Gerrit-Reviewer: Zoltan Chovan <[email protected]>
Gerrit-Comment-Date: Wed, 03 Sep 2025 05:51:33 +0000
Gerrit-HasComments: Yes

Reply via email to