zhuzhurk commented on a change in pull request #9778: [FLINK-14206][runtime]
Let fullRestart metric count both full restarts and fine grained restarts
URL: https://github.com/apache/flink/pull/9778#discussion_r335272545
##########
File path:
flink-runtime/src/main/java/org/apache/flink/runtime/executiongraph/metrics/NumberOfFullRestartsGauge.java
##########
@@ -42,6 +42,6 @@ public NumberOfFullRestartsGauge(ExecutionGraph
executionGraph) {
@Override
public Long getValue() {
- return eg.getNumberOfFullRestarts();
+ return eg.getNumberOfRestarts();
Review comment:
It's a good question. The metric reporting happens in a separate thread
("Flink-MetricRegistry"). So dirty read is possible to happen if a metric
reports value which is modified in another thread.
But I'm not sure whether we need to guarantee there is no dirty read in
metric reporting.
I took a look and find almost all metrics in Flink does not take this into
consideration.
And I guess `ExecutionGraph#globalModVersion` was made to be volatile mainly
for its version functionality.
WDYT?
----------------------------------------------------------------
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]
With regards,
Apache Git Services
