[GitHub] [iceberg] rymurr commented on a change in pull request #2325: Hive: Use HiveClientPool cache instead of HiveCatalog global cache.

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

rymurr commented on a change in pull request #2325:
URL: https://github.com/apache/iceberg/pull/2325#discussion_r593053021



##########
File path: hive-metastore/src/main/java/org/apache/iceberg/hive/HiveTableOperations.java
##########
@@ -98,15 +97,16 @@
   private final long lockCheckMinWaitTime;
   private final long lockCheckMaxWaitTime;
   private final FileIO fileIO;
+  private final HiveCatalog catalog;
 
-  protected HiveTableOperations(Configuration conf, HiveClientPool metaClients, FileIO fileIO,
-                                String catalogName, String database, String table) {
+  protected HiveTableOperations(Configuration conf, FileIO fileIO, HiveCatalog catalog,

Review comment:
       Along w/ my comment above I don't think the table ops should get to see the entire catalog, rather just the client pool

##########
File path: hive-metastore/src/main/java/org/apache/iceberg/hive/HiveCatalog.java
##########
@@ -51,36 +55,48 @@
 import org.apache.iceberg.exceptions.NoSuchTableException;
 import org.apache.iceberg.hadoop.HadoopFileIO;
 import org.apache.iceberg.io.FileIO;
-import org.apache.iceberg.relocated.com.google.common.base.Joiner;
+import org.apache.iceberg.relocated.com.google.common.annotations.VisibleForTesting;
 import org.apache.iceberg.relocated.com.google.common.base.MoreObjects;
 import org.apache.iceberg.relocated.com.google.common.base.Preconditions;
 import org.apache.iceberg.relocated.com.google.common.collect.ImmutableList;
 import org.apache.iceberg.relocated.com.google.common.collect.Maps;
+import org.apache.iceberg.relocated.com.google.common.util.concurrent.ThreadFactoryBuilder;
+import org.apache.iceberg.util.Pair;
 import org.apache.iceberg.util.PropertyUtil;
 import org.apache.thrift.TException;
 import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;
 
-public class HiveCatalog extends BaseMetastoreCatalog implements Closeable, SupportsNamespaces, Configurable {
+public class HiveCatalog extends BaseMetastoreCatalog implements SupportsNamespaces, Configurable {
   private static final Logger LOG = LoggerFactory.getLogger(HiveCatalog.class);
 
+  private static final String CACHE_CLEANER_INTERVAL = "iceberg.hive.client-pool-cache-cleaner-interval";
+  private static final long CACHE_CLEANER_INTERVAL_DEFAULT = TimeUnit.SECONDS.toMillis(30);
+  private static final String CACHE_EVICTION_INTERVAL = "iceberg.hive.client-pool-cache-eviction-interval";
+  private static final long CACHE_EVICTION_INTERVAL_DEFAULT = TimeUnit.MINUTES.toMillis(5);
+
+  @VisibleForTesting
+  static final Cache<String, Pair<HiveClientPool, Long>> CLIENT_POOL_CACHE = Caffeine.newBuilder()

Review comment:
       I agree. From my reading of this PR most of the client pool related stuff is a separate concern to the catalog itself and should be in a separate class.




----------------------------------------------------------------
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



---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@iceberg.apache.org
For additional commands, e-mail: issues-help@iceberg.apache.org