-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/74559/
-----------------------------------------------------------
Review request for atlas, Madhan Neethiraj and Pinal Shah.
Bugs: ATLAS-4787
https://issues.apache.org/jira/browse/ATLAS-4787
Repository: atlas
Description
-------
in purgeMetricsStats() method Atlas first queries all AtlasMetricsStat entries,
filters out those who passed their TTL, and then requeryies the graph backend
again for each entry by collectionTime just to retrieve guid which was already
available beforehand. This is an n+1 error and completely unnecessary. This
also causes an issue if atlas.cluster.name property was changed after some time
(which may happen in staging environments) and then deleting by collectionTime
of entries before the cluster.name change becomes impossible and yields an
AtlasBaseException
org.apache.atlas.exception.AtlasBaseException: Instance __AtlasMetricsStat with
unique attribute {metricsId=atlas_metrics_1671346802778@newclustername} does
not exist
which then interrupts entire transaction and blocks purging of obsolete entries.
Diffs
-----
repository/src/main/java/org/apache/atlas/services/MetricsService.java
9ec2cd2cb
Diff: https://reviews.apache.org/r/74559/diff/1/
Testing
-------
Tested on my cluster and run mvn clean test
Thanks,
Szymon Orzechowski
