dongjoon-hyun commented on a change in pull request #32388:
URL: https://github.com/apache/spark/pull/32388#discussion_r642167700



##########
File path: 
common/network-shuffle/src/main/java/org/apache/spark/network/shuffle/ExternalBlockHandler.java
##########
@@ -264,6 +265,8 @@ private void checkAuth(TransportClient client, String 
appId) {
     private final Timer registerExecutorRequestLatencyMillis = new Timer();
     // Time latency for processing finalize shuffle merge request latency in ms
     private final Timer finalizeShuffleMergeLatencyMillis = new Timer();
+    // Block transfer rate in blocks per second

Review comment:
       Is this valid when we do `getContinuousBlocksData`?

##########
File path: 
common/network-yarn/src/main/java/org/apache/spark/network/yarn/YarnShuffleServiceMetrics.java
##########
@@ -78,7 +82,32 @@ public static void collectMetric(
           new ShuffleServiceMetricsInfo(name + "_rate1", "1 minute rate of 
timer " + name),
           t.getOneMinuteRate())
         .addGauge(new ShuffleServiceMetricsInfo(name + "_rateMean", "Mean rate 
of timer " + name),
-          t.getMeanRate());
+          t.getMeanRate())
+        .addGauge(
+          getShuffleServiceMetricsInfoForGenericValue(timingName, "max"), 
snapshot.getMax())
+        .addGauge(
+          getShuffleServiceMetricsInfoForGenericValue(timingName, "min"), 
snapshot.getMin())
+        .addGauge(
+          getShuffleServiceMetricsInfoForGenericValue(timingName, "mean"), 
snapshot.getMean())
+        .addGauge(
+          getShuffleServiceMetricsInfoForGenericValue(timingName, "stdDev"), 
snapshot.getStdDev());
+      for (int percentileThousands : new int[] { 10, 50, 250, 500, 750, 950, 
980, 990, 999 }) {
+        String percentileStr;
+        switch (percentileThousands) {
+          case 10:
+            percentileStr = "1stPercentile";
+            break;
+          case 999:
+            percentileStr = "99.9thPercentile";

Review comment:
       Hi, @xkrogen . Thank you for adding this and I saw your 
[comment](https://github.com/apache/spark/pull/32388#discussion_r629507805).
   
   Technically, `999thPercentile` is consistent with the conventional in Apache 
Spark instead of `99.9thPercentitle`. If you look at the other existing 
metrics, it has the following.
   <img width="146" alt="Screen Shot 2021-05-30 at 6 52 14 PM" 
src="https://user-images.githubusercontent.com/9700541/120129072-325b1b80-c178-11eb-9ff2-dd242951e24c.png";>
   
   
   And, just a question. May I ask where you borrow this `99.9thPercentitle`?




-- 
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]



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

Reply via email to