d-wang-commit commented on code in PR #18692:
URL: https://github.com/apache/iotdb/pull/18692#discussion_r4078528246


##########
iotdb-core/confignode/src/main/java/org/apache/iotdb/confignode/manager/partition/PartitionManager.java:
##########
@@ -645,16 +651,62 @@ private TSStatus extendRegionGroupIfNecessary(
     return result;
   }
 
+  private TSStatus proactiveExtendRegionGroupIfNecessary(
+      final Map<String, ? extends Collection<TSeriesPartitionSlot>> 
unassignedPartitionSlotsMap,
+      final TConsensusGroupType consensusGroupType)
+      throws DatabaseNotExistsException, NotEnoughDataNodeException {
+    final Map<String, Integer> allotmentMap = new HashMap<>();
+    for (Map.Entry<String, ? extends Collection<TSeriesPartitionSlot>> entry :
+        unassignedPartitionSlotsMap.entrySet()) {
+      final String database = entry.getKey();
+      final int maxRegionGroupCount =
+          getClusterSchemaManager().getMaxRegionGroupNum(database, 
consensusGroupType);
+      final int minRegionGroupCount =
+          getClusterSchemaManager().getMinRegionGroupNum(database, 
consensusGroupType);
+      final int allocatedRegionGroupCount =
+          partitionInfo.getRegionGroupCount(database, consensusGroupType);
+      // As with AUTO, grow toward the minimum by at most the number of 
pending series slots.
+      // An existing data slot with new time partitions also contributes to 
this incremental growth.
+      final int minimumRegionGroupTarget =
+          Math.min(
+              minRegionGroupCount,
+              allocatedRegionGroupCount + new 
HashSet<>(entry.getValue()).size());
+      final int targetRegionGroupCount =
+          Math.min(
+              maxRegionGroupCount,
+              Math.max(
+                  minimumRegionGroupTarget,
+                  partitionInfo.getSeriesPartitionSlotsCount(
+                      database, consensusGroupType, entry.getValue())));
+      // Also allocate one group per active series slot, up to the same 
maximum as AUTO.
+      // Existing groups are retained when the maximum decreases or the policy 
changes.
+      if (allocatedRegionGroupCount < targetRegionGroupCount) {
+        allotmentMap.put(database, targetRegionGroupCount - 
allocatedRegionGroupCount);
+      } else if (!entry.getValue().isEmpty()
+          && allocatedRegionGroupCount > 0
+          && allocatedRegionGroupCount < maxRegionGroupCount
+          && partitionInfo.getAllRegionGroupIds(database, 
consensusGroupType).stream()
+              .allMatch(
+                  regionGroupId ->
+                      RegionGroupStatus.Disabled.equals(
+                          
getLoadManager().getRegionGroupStatus(regionGroupId)))) {

Review Comment:
   The max region group number caps any extra allocation. Also, a successfully 
created group is immediately activated, so subsequent requests no longer 
satisfy the “all disabled” condition. Given the bounded impact, we’ll keep the 
current implementation



-- 
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.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to