[fix](schemeChange) fe oom because replicas too many when schema change

docs/zh-CN/docs/admin-manual/config/fe-config.md

@@ -2313,3 +2313,14 @@ load 标签清理器将每隔 `label_clean_interval_second` 运行一次以清
 是否可以动态配置：true
 
 是否为 Master FE 节点独有的配置项：true
+
+
+### max_replica_count_when_schema_change
+
+OlapTable在做schema change时，允许的最大副本数，副本数过大会导致FE OOM。
+
+默认值：100000
+
+是否可以动态配置：true
+
+是否为 Master FE 节点独有的配置项：true

fe/fe-core/src/main/java/org/apache/doris/alter/AlterHandler.java

@@ -27,6 +27,7 @@
 import org.apache.doris.catalog.Replica;
 import org.apache.doris.catalog.Table;
 import org.apache.doris.catalog.Tablet;
+import org.apache.doris.common.Config;
 import org.apache.doris.common.DdlException;
 import org.apache.doris.common.FeConstants;
 import org.apache.doris.common.MetaNotFoundException;
@@ -253,4 +254,23 @@ public void replayAlterJobV2(AlterJobV2 alterJob) {
             existingJob.replay(alterJob);
         }
     }
+
+    /**
+     * there will be OOM if there are too many replicas of the table when schema change.
+     */
+    protected void checkReplicaCount(OlapTable olapTable) throws DdlException {
+        olapTable.readLock();
+        try {
+            long replicaCount = olapTable.getReplicaCount();
+            long maxReplicaCount = Config.max_replica_count_when_schema_change;
+            if (replicaCount > maxReplicaCount) {
+                String msg = String.format("%s have %d replicas reach %d limit when schema change.",
+                        olapTable.getName(), replicaCount, maxReplicaCount);
+                LOG.warn(msg);
+                throw new DdlException(msg);
+            }
+        } finally {
+            olapTable.readUnlock();
+        }
+    }
 }

fe/fe-core/src/main/java/org/apache/doris/alter/MaterializedViewHandler.java

@@ -233,6 +233,7 @@ public void processCreateMaterializedView(CreateMaterializedViewStmt addMVClause
      */
     public void processBatchAddRollup(List<AlterClause> alterClauses, Database db, OlapTable olapTable)
             throws DdlException, AnalysisException {
+        checkReplicaCount(olapTable);
         Map<String, RollupJobV2> rollupNameJobMap = new LinkedHashMap<>();
         // save job id for log
         Set<Long> logJobIdSet = new HashSet<>();

fe/fe-core/src/main/java/org/apache/doris/alter/SchemaChangeHandler.java

@@ -1139,6 +1139,7 @@ private void createJob(long dbId, OlapTable olapTable, Map<Long, LinkedList<Colu
             throw new DdlException("Table[" + olapTable.getName() + "]'s is doing ROLLUP job");
         }
 
+        checkReplicaCount(olapTable);
         // for now table's state can only be NORMAL
         Preconditions.checkState(olapTable.getState() == OlapTableState.NORMAL, olapTable.getState().name());
 

fe/fe-core/src/main/java/org/apache/doris/common/Config.java

@@ -1831,4 +1831,12 @@ public class Config extends ConfigBase {
      */
     @ConfField(mutable = true, masterOnly = false)
     public static long hive_metastore_client_timeout_second = 10;
+
+    /**
+     * if table has too many replicas, Fe occur oom when schema change.
+     * 10W replicas is a reasonable value for testing.
+     */
+    @ConfField(mutable = true, masterOnly = true)
+    public static long max_replica_count_when_schema_change = 100000;
 }
+