[ 
https://issues.apache.org/jira/browse/FLINK-5090?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15922825#comment-15922825
 ] 

ASF GitHub Bot commented on FLINK-5090:
---------------------------------------

Github user zentol commented on a diff in the pull request:

    https://github.com/apache/flink/pull/3348#discussion_r105753418
  
    --- Diff: 
flink-runtime/src/main/java/org/apache/flink/runtime/io/network/partition/consumer/InputGateMetrics.java
 ---
    @@ -0,0 +1,168 @@
    +/*
    + * 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.flink.runtime.io.network.partition.consumer;
    +
    +import org.apache.flink.metrics.Gauge;
    +import org.apache.flink.metrics.MetricGroup;
    +
    +import static org.apache.flink.util.Preconditions.checkNotNull;
    +
    +public class InputGateMetrics {
    +
    +   private final SingleInputGate inputGate;
    +
    +   private long lastTotal = -1;
    +
    +   private int lastMin = -1;
    +
    +   private int lastMax = -1;
    +
    +   private float lastAvg = -1.0f;
    +
    +   // 
------------------------------------------------------------------------
    +
    +   private InputGateMetrics(SingleInputGate inputGate) {
    +           this.inputGate = checkNotNull(inputGate);
    +   }
    +
    +   // 
------------------------------------------------------------------------
    +
    +   // these methods are package private to make access from the nested 
classes faster 
    +
    +   long refreshAndGetTotal() {
    +           long total;
    +           if ((total = lastTotal) == -1) {
    +                   refresh();
    --- End diff --
    
    Custom objects can't be displayed properly in the web interface since we 
call ```toString()``` on it. The same happens in most reporters; so this isn't 
really an option.
    
    As it stands we don't have a single metrics that is guaranteed to be 100% 
consistent with other metrics. numRecordsOut and numBytesOut to not descriibe 
the same moment in time. Neither is this guaranteed for the checkpoint metrics; 
while these are updated all at once (from the outside), there is no mechanism 
that prevents this update in the middle of a report.
    
    I don't know a lot about the network stack; so whether it is truly 
necessary to have all metrics describe one point in time I can't say.
    
    If this is necessary the only way i can think of right now is abusing the 
View metric type. View's are meant an add-on for metrics that want to be 
updated in regular intervals (5 seconds) regardless of when their value is 
actually requested. A metric that only implements the View interface is never 
reported, but still updated, so you could have this view update a shared 
data-structure from which the other gauges simply retrieve the current value,
    
    If this is not necessary i would simply separate them and don't worry about 
the performance overhead of the metrics; as long as this doesn't affect the job 
via taking locks or similar.



> Expose optionally detailed metrics about network queue lengths
> --------------------------------------------------------------
>
>                 Key: FLINK-5090
>                 URL: https://issues.apache.org/jira/browse/FLINK-5090
>             Project: Flink
>          Issue Type: New Feature
>          Components: Metrics, Network
>    Affects Versions: 1.1.3
>            Reporter: Stephan Ewen
>            Assignee: Stephan Ewen
>
> For debugging purposes, it is important to have access to more detailed 
> metrics about the length of network input and output queues.



--
This message was sent by Atlassian JIRA
(v6.3.15#6346)

Reply via email to