jiajunwang commented on a change in pull request #639: Refine the WAGED
rebalancer to minimize the partial rebalance workload.
URL: https://github.com/apache/helix/pull/639#discussion_r352866188
##########
File path:
helix-core/src/main/java/org/apache/helix/controller/rebalancer/waged/WagedRebalancer.java
##########
@@ -389,50 +376,81 @@ public void close() {
return finalIdealStateMap;
}
- // TODO make the Baseline calculation async if complicated algorithm is used
for the Baseline
- private void refreshBaseline(ResourceControllerDataProvider clusterData,
- Map<HelixConstants.ChangeType, Set<String>> clusterChanges, Map<String,
Resource> resourceMap,
- final CurrentStateOutput currentStateOutput)
+ private void globalRebalance(ResourceControllerDataProvider clusterData,
+ Map<String, Resource> resourceMap, final CurrentStateOutput
currentStateOutput,
+ RebalanceAlgorithm algorithm)
throws HelixRebalanceException {
+ _changeDetector.updateSnapshots(clusterData);
+ // Get all the changed items' information. Filter for the items that have
content changed.
+ final Map<HelixConstants.ChangeType, Set<String>> clusterChanges =
+ _changeDetector.getChangeTypes().stream()
+ .collect(Collectors.toMap(changeType -> changeType, changeType -> {
+ Set<String> itemKeys = new HashSet<>();
+ itemKeys.addAll(_changeDetector.getAdditionsByType(changeType));
+ itemKeys.addAll(_changeDetector.getChangesByType(changeType));
+ itemKeys.addAll(_changeDetector.getRemovalsByType(changeType));
+ return itemKeys;
+ })).entrySet().stream().filter(changeEntry ->
!changeEntry.getValue().isEmpty())
+ .collect(Collectors
+ .toMap(changeEntry -> changeEntry.getKey(), changeEntry ->
changeEntry.getValue()));
+
if (clusterChanges.keySet().stream()
.anyMatch(GLOBAL_REBALANCE_REQUIRED_CHANGE_TYPES::contains)) {
- LOG.info("Start calculating the new baseline.");
- _globalBaselineCalcCounter.increment(1L);
- _globalBaselineCalcLatency.startMeasuringLatency();
-
- // For baseline calculation
+ // Build the cluster model for rebalance calculation.
+ // Note, for a baseline calculation,
Review comment:
I can change it but it might cause more confusion because the following
logic is really calculating a new baseline. And the methods are all called
something related to the baseline. So prefer to keep this description.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
With regards,
Apache Git Services
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
