[
https://issues.apache.org/jira/browse/GOBBLIN-1657?focusedWorklogId=778814&page=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-778814
]
ASF GitHub Bot logged work on GOBBLIN-1657:
-------------------------------------------
Author: ASF GitHub Bot
Created on: 07/Jun/22 02:00
Start Date: 07/Jun/22 02:00
Worklog Time Spent: 10m
Work Description: vikrambohra commented on code in PR #3517:
URL: https://github.com/apache/gobblin/pull/3517#discussion_r890699926
##########
gobblin-completeness/src/main/java/org/apache/gobblin/completeness/verifier/KafkaAuditCountVerifier.java:
##########
@@ -45,6 +45,8 @@ public class KafkaAuditCountVerifier {
public static final String REFERENCE_TIERS = COMPLETENESS_PREFIX +
"reference.tiers";
public static final String THRESHOLD = COMPLETENESS_PREFIX + "threshold";
private static final double DEFAULT_THRESHOLD = 0.999;
+ public static final String COMPLETE_ON_NO_COUNTS = COMPLETENESS_PREFIX +
"complete.on.no.counts";
Review Comment:
Default behavior is for other users of this class to not want a "success" if
there is an empty response from kafka audit.
This config is only for this use case where we want a "success" on empty
response from kafka audit.
Issue Time Tracking
-------------------
Worklog Id: (was: 778814)
Time Spent: 1h 10m (was: 1h)
> Update completion watermark on change_property in IcebergMetadataWriter
> ------------------------------------------------------------------------
>
> Key: GOBBLIN-1657
> URL: https://issues.apache.org/jira/browse/GOBBLIN-1657
> Project: Apache Gobblin
> Issue Type: Task
> Reporter: Vikram Bohra
> Priority: Major
> Time Spent: 1h 10m
> Remaining Estimate: 0h
>
> There are quiet topics where data can be generated quiet infrequently causing
> the completion watermark to lag. Also, a commit operation could miss
> completion watermark update if there is a lag in kafka audit.
> To fix above issues:
> # Added logic to check audit count on change_property operation type for the
> next window of current watermark
> # If counts match or an empty map is returned then consider audit count as
> complete and update the watermark.
> Added a config in KafkaAuditVerifier to return complete if found an empty map
> (with no counts from audit system)
--
This message was sent by Atlassian Jira
(v8.20.7#820007)
