[
https://issues.apache.org/jira/browse/ATLAS-4787?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17757917#comment-17757917
]
Szymon Orzechowski commented on ATLAS-4787:
-------------------------------------------
Review board: https://reviews.apache.org/r/74559/
> MetricsService.purgeMetricsStats() does unnecessary queries to graph backend
> ----------------------------------------------------------------------------
>
> Key: ATLAS-4787
> URL: https://issues.apache.org/jira/browse/ATLAS-4787
> Project: Atlas
> Issue Type: Bug
> Components: atlas-core
> Affects Versions: 3.0.0, 2.3.0
> Reporter: Szymon Orzechowski
> Priority: Major
> Attachments: ATLAS-4787.patch
>
>
> As of now, current code of MetricsService.purgeMetricsStats():
> # retrieves all metric stats (getAllMetricsStats(true))
> # filters out metrics to delete
> # iteraters over each entry (deleteMetricsStatByCollectionTime)
> ## queries metric stat by timestamp
> ## deletes metric stat by guid
> This however is an n+1 error because at step 1. we already have the necessary
> guids. At step 3. Atlas could either do a batch delete or delete by guid
> right off the bat.
>
> Deleting by collectionTime also causes an issue if atlas.cluster.name was
> changed after some time, because it would try to look for obsolete metric
> stat with new cluster name, which results with
> {code:java}
> org.apache.atlas.exception.AtlasBaseException: Instance __AtlasMetricsStat
> with unique attribute {metricsId=atlas_metrics_1671346802778@newclustername}
> does not exist
> {code}
> and rollbacks the entire transaction.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
