deniskuzZ commented on code in PR #6379:
URL: https://github.com/apache/hive/pull/6379#discussion_r3361970573


##########
ql/src/java/org/apache/hadoop/hive/llap/LlapHiveUtils.java:
##########
@@ -74,18 +75,21 @@ public static PartitionDesc partitionDescForPath(Path path, 
Map<Path, PartitionD
     return part;
   }
 
-  public static CacheTag getDbAndTableNameForMetrics(Path path, boolean 
includeParts,
-      PartitionDesc part) {
-
-    // Fallback to legacy cache tag creation logic.
+  /**
+   * Builds a {@link CacheTag} for the given path and partition descriptor.
+   * The catalog name is derived from the {@link PartitionDesc} when 
available, falling back
+   * to {@link Warehouse#DEFAULT_CATALOG_NAME} when {@code part} is null.
+   */
+  public static CacheTag getCacheTag(Path path, boolean includeParts, 
PartitionDesc part) {
     if (part == null) {
-      return CacheTag.build(LlapUtil.getDbAndTableNameForMetrics(path, 
includeParts));
+      return CacheTag.build(
+          Warehouse.DEFAULT_CATALOG_NAME, 
LlapUtil.getDbAndTableNameForMetrics(path, includeParts));

Review Comment:
   Job configuration probably won't work, we might have query involving 
multiple catalogs... 
   Let's stick with default
   ````
   if (part == null) {
       // No table metadata on this read path and the path doesn't encode the 
catalog, so the
       // catalog can't be determined authoritatively. Returning null is not an 
option: when
       // LLAP_TRACK_CACHE_USAGE is on (the only time we get here) the buffer 
MUST carry a non-null
       // tag — CacheContentsTracker keys buffers by tag in a 
ConcurrentSkipListMap, and the
       // proactive-eviction predicate dereferences the tag without a null 
check. Fall back to the
       // default catalog; this is best-effort and effectively unreachable for 
real cache-tracked scans.
       return CacheTag.build(
           Warehouse.DEFAULT_CATALOG_NAME, 
LlapUtil.getDbAndTableNameForMetrics(path, includeParts));
     }
   ````



-- 
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]


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to