[HUDI-4814] Schedules new clustering plan based on latest clustering instant #6574
There are no files selected for viewing
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
|
|
@@ -48,21 +48,28 @@ public class ClusteringPlanActionExecutor<T extends HoodieRecordPayload, I, K, O | |
| private final Option<Map<String, String>> extraMetadata; | ||
|
|
||
| public ClusteringPlanActionExecutor(HoodieEngineContext context, | ||
| HoodieWriteConfig config, | ||
| HoodieTable<T, I, K, O> table, | ||
| String instantTime, | ||
| Option<Map<String, String>> extraMetadata) { | ||
| super(context, config, table, instantTime); | ||
| this.extraMetadata = extraMetadata; | ||
| } | ||
|
|
||
| protected Option<HoodieClusteringPlan> createClusteringPlan() { | ||
| LOG.info("Checking if clustering needs to be run on " + config.getBasePath()); | ||
| Option<HoodieInstant> lastClusteringInstant; | ||
| Option<HoodieInstant> pendingInstant = table.getActiveTimeline().filterPendingReplaceTimeline().lastInstant(); | ||
| if (pendingInstant.isPresent()) { | ||
| lastClusteringInstant = pendingInstant; | ||
| } else { | ||
| lastClusteringInstant = table.getActiveTimeline().getCompletedReplaceTimeline().lastInstant(); | ||
| } | ||
|
|
||
| int commitsSinceLastClustering = table.getActiveTimeline().getCommitsTimeline().filterCompletedInstants() | ||
| .findInstantsAfter(lastClusteringInstant.map(HoodieInstant::getTimestamp).orElse("0"), Integer.MAX_VALUE) | ||
| .countInstants(); | ||
|
|
||
| if (config.inlineClusteringEnabled() && config.getInlineClusterMaxCommits() > commitsSinceLastClustering) { | ||
|
Contributor
There was a problem hiding this comment. This and the condition below guarantee that the clustering is only scheduled based on the max_commits config. @eric9204 could you double check the logic?
Contributor
Author
There was a problem hiding this comment. @yihua yes, this is indeed a redundant inspection, I'm testing whether this condition is needed. By adding these two conditions, it can really be guaranteed that only one clustering is running at the same time, and if there is no completed clustering, no new clustering plan will be generated. Configure only these three parameters. 'clustering.schedule.enabled'='true', Configure only these three parameters. 'clustering.schedule.enabled'='true',
Contributor
There was a problem hiding this comment. ok, so the issue we are trying to solve is: there is a regular writer which just schedules clustering and we have a async clustering job which does the execution of clustering. if clustering is pending (may be will be executed by an async clustering job), every new successful commit with regular writer will keep adding new replacecommit.requested. If yes, then the fix makes sense to me.
Contributor
There was a problem hiding this comment. but one thing which I am finding it hard to comprehend is. wrt clustering, either both planning and execution is inline. or both are async atleast wrt spark datasource writer. So, not sure how the user ended up where clustering was just scheduled w/o getting to completion.
Contributor
There was a problem hiding this comment. Yes, the fix makes sense to me too.
Flink writer schedules a clustering plan on each successful regular commit and there is a async pipeline that executes the clustering continuously, this patch can solve the problem that the clustering plan schedules too frequently if there is pending clustering. So, +1 from my side. |
||
| LOG.info("Not scheduling inline clustering as only " + commitsSinceLastClustering | ||
| + " commits was found since last clustering " + lastClusteringInstant + ". Waiting for " | ||
|
|
||
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
guess we can filter the instants directly by the action type here ?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
@danny0405 good idea,I'm working on it.