gianm closed pull request #5974: [Backport] Coordinator fix balance to try to move max segments instead of up to max segments URL: https://github.com/apache/incubator-druid/pull/5974
This is a PR merged from a forked repository. As GitHub hides the original diff on merge, it is displayed below for the sake of provenance: As this is a foreign pull request (from a fork), the diff is supplied below (as it won't show otherwise due to GitHub magic): diff --git a/server/src/main/java/io/druid/server/coordinator/helper/DruidCoordinatorBalancer.java b/server/src/main/java/io/druid/server/coordinator/helper/DruidCoordinatorBalancer.java index cc266d705df..b4f7e22411f 100644 --- a/server/src/main/java/io/druid/server/coordinator/helper/DruidCoordinatorBalancer.java +++ b/server/src/main/java/io/druid/server/coordinator/helper/DruidCoordinatorBalancer.java @@ -20,9 +20,9 @@ package io.druid.server.coordinator.helper; import com.google.common.collect.Lists; -import io.druid.java.util.emitter.EmittingLogger; import io.druid.client.ImmutableDruidServer; import io.druid.java.util.common.StringUtils; +import io.druid.java.util.emitter.EmittingLogger; import io.druid.server.coordinator.BalancerSegmentHolder; import io.druid.server.coordinator.BalancerStrategy; import io.druid.server.coordinator.CoordinatorStats; @@ -40,6 +40,7 @@ import java.util.NavigableSet; import java.util.SortedSet; import java.util.concurrent.ConcurrentHashMap; +import java.util.stream.Collectors; /** */ @@ -103,15 +104,16 @@ private void balanceTier( return; } - final List<ServerHolder> serverHolderList = Lists.newArrayList(servers); + final List<ServerHolder> toMoveFrom = Lists.newArrayList(servers); + final List<ServerHolder> toMoveTo = Lists.newArrayList(servers); - if (serverHolderList.size() <= 1) { + if (toMoveTo.size() <= 1) { log.info("[%s]: One or fewer servers found. Cannot balance.", tier); return; } int numSegments = 0; - for (ServerHolder server : serverHolderList) { + for (ServerHolder server : toMoveFrom) { numSegments += server.getServer().getSegments().size(); } @@ -119,20 +121,31 @@ private void balanceTier( log.info("No segments found. Cannot balance."); return; } + + final int maxToLoad = params.getCoordinatorDynamicConfig().getMaxSegmentsInNodeLoadingQueue(); long unmoved = 0L; - for (int iter = 0; iter < maxSegmentsToMove; iter++) { - final BalancerSegmentHolder segmentToMove = strategy.pickSegmentToMove(serverHolderList); + for (int moved = 0; (moved + unmoved) < maxSegmentsToMove;) { + final BalancerSegmentHolder segmentToMove = strategy.pickSegmentToMove(toMoveFrom); if (segmentToMove != null && params.getAvailableSegments().contains(segmentToMove.getSegment())) { - final ServerHolder holder = strategy.findNewSegmentHomeBalancer(segmentToMove.getSegment(), serverHolderList); + final List<ServerHolder> toMoveToWithLoadQueueCapacity = + toMoveTo.stream() + .filter(s -> maxToLoad <= 0 || s.getNumberOfSegmentsInQueue() < maxToLoad) + .collect(Collectors.toList()); - if (holder != null) { - moveSegment(segmentToMove, holder.getServer(), params); + final ServerHolder destinationHolder = + strategy.findNewSegmentHomeBalancer(segmentToMove.getSegment(), toMoveToWithLoadQueueCapacity); + + if (destinationHolder != null) { + moveSegment(segmentToMove, destinationHolder.getServer(), params); + moved++; } else { - ++unmoved; + log.info("Segment [%s] is 'optimally' placed.", segmentToMove.getSegment().getIdentifier()); + unmoved++; } } } + if (unmoved == maxSegmentsToMove) { // Cluster should be alive and constantly adjusting log.info("No good moves found in tier [%s]", tier); @@ -140,7 +153,7 @@ private void balanceTier( stats.addToTieredStat("unmovedCount", tier, unmoved); stats.addToTieredStat("movedCount", tier, currentlyMovingSegments.get(tier).size()); if (params.getCoordinatorDynamicConfig().emitBalancingStats()) { - strategy.emitStats(tier, stats, serverHolderList); + strategy.emitStats(tier, stats, toMoveFrom); } log.info( "[%s]: Segments Moved: [%d] Segments Let Alone: [%d]", ---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@druid.apache.org For additional commands, e-mail: dev-h...@druid.apache.org