mcvsubbu commented on a change in pull request #4156: Refactor 
HelixExternalViewBasedTimeBoundaryService to support all time units
URL: https://github.com/apache/incubator-pinot/pull/4156#discussion_r279950299
 
 

 ##########
 File path: 
pinot-broker/src/main/java/org/apache/pinot/broker/routing/HelixExternalViewBasedTimeBoundaryService.java
 ##########
 @@ -48,77 +50,106 @@ public 
HelixExternalViewBasedTimeBoundaryService(ZkHelixPropertyStore<ZNRecord>
   }
 
   public void updateTimeBoundaryService(ExternalView externalView) {
-    if (_propertyStore == null) {
-      return;
-    }
-    String tableName = externalView.getResourceName();
-    // Do nothing for realtime table.
-    if (TableNameBuilder.getTableTypeFromTableName(tableName) == 
TableType.REALTIME) {
+    String tableNameWithType = externalView.getResourceName();
+
+    // Skip real-time table, only use offline table to update the time boundary
+    if (TableNameBuilder.getTableTypeFromTableName(tableNameWithType) == 
TableType.REALTIME) {
       return;
     }
 
     Set<String> offlineSegmentsServing = externalView.getPartitionSet();
     if (offlineSegmentsServing.isEmpty()) {
-      LOGGER.info("Skipping updating time boundary service for table '{}' with 
no offline segments.", tableName);
+      LOGGER.warn("Skipping updating time boundary for table: '{}' with no 
offline segment", tableNameWithType);
       return;
     }
 
-    TableConfig offlineTableConfig = 
ZKMetadataProvider.getOfflineTableConfig(_propertyStore, tableName);
-    assert offlineTableConfig != null;
-    TimeUnit tableTimeUnit = 
offlineTableConfig.getValidationConfig().getTimeType();
-    if (tableTimeUnit == null) {
-      LOGGER.info("Skipping updating time boundary service for table '{}' 
because time unit is not set", tableName);
+    // TODO: when we start using dateTime, pick the time column from the 
retention config, and use the DateTimeFieldSpec
+    //       from the schema to determine the time unit
+    // TODO: support SDF
+    TableConfig tableConfig = 
ZKMetadataProvider.getTableConfig(_propertyStore, tableNameWithType);
+    assert tableConfig != null;
+    SegmentsValidationAndRetentionConfig retentionConfig = 
tableConfig.getValidationConfig();
+    String timeColumn = retentionConfig.getTimeColumnName();
+    TimeUnit tableTimeUnit = retentionConfig.getTimeType();
+    if (timeColumn == null || tableTimeUnit == null) {
+      LOGGER.error("Skipping updating time boundary for table: '{}' because 
time column/unit is not set",
+          tableNameWithType);
       return;
     }
 
-    // Bulk reading all segment zk-metadata at once is more efficient than 
reading one at a time.
+    Schema schema = ZKMetadataProvider.getTableSchema(_propertyStore, 
tableNameWithType);
 
 Review comment:
   Why do the schema check here? We should be doing that when schema gets 
updated, (and also when tableconfig gets updated). Are you trying to protect 
existing use cases that may have inconsistent time column definitions? Perhaps 
running a one-time audit will be better in that case.
   
   We will be reading the schema for each EV update -- each time retention 
manager removes a segment, or a new realtime segment gets added.

----------------------------------------------------------------
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]

Reply via email to