steveloughran commented on code in PR #1187:
URL: https://github.com/apache/parquet-mr/pull/1187#discussion_r1429822224


##########
parquet-hadoop/src/main/java/org/apache/parquet/hadoop/ParquetMetricsCallback.java:
##########
@@ -0,0 +1,34 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements.  See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership.  The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License.  You may obtain a copy of the License at
+ *
+ *   http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing,
+ * software distributed under the License is distributed on an
+ * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+ * KIND, either express or implied.  See the License for the
+ * specific language governing permissions and limitations
+ * under the License.
+ */
+package org.apache.parquet.hadoop;
+
+/**
+ *  a simple interface to pass basic metric values by name to any 
implementation. Typically an
+ *  implementation of this interface will serve as a bridge to pass metric 
values on
+ *  to the metrics system of a distributed engine (hadoop, spark, etc).
+ */
+public interface ParquetMetricsCallback {

Review Comment:
   Can you tag as unstable but promise that all new callbacks will have a 
default implementation for any new metric which will be a no-op, eg
   ```
   default addMaximum(key, value) { } ;
   ```
   
   I'd also propose a specific "addDuration(key, duration)" call. For iostats 
we do min/mean/max and sum values here as they are all relevant
   
   Note also that the nanotimer is a tricky one. It is low cost and required to 
go monotonincally forward on a single core -but that doesn't hold across 
sockets. Older Intel parts don't even do this on cores on the same die. Not 
saying don't use it, but be prepared for the diff to be negative and if so set 
to zero.
   For this reason, and because exceptions may be a very different duration 
category from successes, consider a DurationTracker class or interface/class 
which hides more of the start and stop timestamp collection and can even be 
told of a failure/ + probed for success vs failure. Then use this as the 
addDuration() value. 



-- 
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