[GitHub] [hbase] saintstack commented on a change in pull request #1786: HBASE-24418 Consolidate Normalizer implementations

[GitBox <gi...@apache.org>]

saintstack commented on a change in pull request #1786:
URL: https://github.com/apache/hbase/pull/1786#discussion_r431457649



##########
File path: hbase-server/src/main/java/org/apache/hadoop/hbase/master/normalizer/SplitPlanFirstComparator.java
##########
@@ -0,0 +1,40 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements.  See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership.  The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License.  You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.hadoop.hbase.master.normalizer;
+
+import java.util.Comparator;
+
+/**
+ * Comparator class that gives higher priority to {@link SplitNormalizationPlan}.
+ */
+class SplitPlanFirstComparator implements Comparator<NormalizationPlan> {

Review comment:
       I meant regions in the table order... i.e. do the plan for early table REgions first..... But yeah, can be punted to later.

##########
File path: hbase-server/src/main/java/org/apache/hadoop/hbase/master/HMaster.java
##########
@@ -1911,43 +1912,51 @@ public boolean normalizeRegions() throws IOException {
       return false;
     }
 
-    synchronized (this.normalizer) {
+    if (!normalizationInProgressLock.tryLock()) {
       // Don't run the normalizer concurrently
+      LOG.info("Normalization already in progress. Skipping request.");
+    } else {
+      try {
+        List<TableName> allEnabledTables = new ArrayList<>(
+          tableStateManager.getTablesInStates(TableState.State.ENABLED));
+        Collections.shuffle(allEnabledTables);
 
-      List<TableName> allEnabledTables = new ArrayList<>(
-        this.tableStateManager.getTablesInStates(TableState.State.ENABLED));
-
-      Collections.shuffle(allEnabledTables);
-
-      for (TableName table : allEnabledTables) {
-        TableDescriptor tblDesc = getTableDescriptors().get(table);
-        if (table.isSystemTable() || (tblDesc != null &&
-            !tblDesc.isNormalizationEnabled())) {
-          LOG.trace("Skipping normalization for {}, as it's either system"
-              + " table or doesn't have auto normalization turned on", table);
-          continue;
-        }
+        for (TableName table : allEnabledTables) {
+          if (table.isSystemTable()) {
+            continue;
+          }
+          final TableDescriptor tblDesc = getTableDescriptors().get(table);
+          if (tblDesc != null && !tblDesc.isNormalizationEnabled()) {
+            LOG.debug(
+              "Skipping {} because normalization is disabled in its table properties.", table);
+            continue;
+          }
 
-        // make one last check that the cluster isn't shutting down before proceeding.
-        if (skipRegionManagementAction("region normalizer")) {
-          return false;
-        }
+          // make one last check that the cluster isn't shutting down before proceeding.
+          if (skipRegionManagementAction("region normalizer")) {
+            return false;
+          }
 
-        final List<NormalizationPlan> plans = this.normalizer.computePlanForTable(table);
-        if (CollectionUtils.isEmpty(plans)) {
-          return true;
-        }
+          final List<NormalizationPlan> plans = normalizer.computePlansForTable(table);
+          if (CollectionUtils.isEmpty(plans)) {
+            return true;
+          }
 
-        try (final Admin admin = asyncClusterConnection.toConnection().getAdmin()) {
-          for (NormalizationPlan plan : plans) {
-            plan.execute(admin);
-            if (plan.getType() == PlanType.SPLIT) {
-              splitPlanCount++;
-            } else if (plan.getType() == PlanType.MERGE) {
-              mergePlanCount++;
+          try (final Admin admin = asyncClusterConnection.toConnection().getAdmin()) {
+            // as of this writing, `plan.execute()` is non-blocking, so there's no artificial rate-
+            // limiting of merge requests due to this serial loop.

Review comment:
       Follow-on or just wait till this an issue. I'd be more concerned about what I think of as too-aggressive behavior.... asking for a split at the S * 2 threshold...




----------------------------------------------------------------
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:
users@infra.apache.org