Abacn commented on code in PR #33043:
URL: https://github.com/apache/beam/pull/33043#discussion_r1878348126
##########
runners/core-java/src/main/java/org/apache/beam/runners/core/metrics/MetricUpdates.java:
##########
@@ -66,21 +68,30 @@ public static <T> MetricUpdate<T> create(MetricKey key, T
update) {
/** All the sets updates. */
public abstract Iterable<MetricUpdate<StringSetData>> stringSetUpdates();
+ /** All the histogram updates. */
+ public abstract Iterable<MetricUpdate<HistogramData>>
perWorkerHistogramsUpdates();
+
/** Create a new {@link MetricUpdates} bundle. */
public static MetricUpdates create(
Iterable<MetricUpdate<Long>> counterUpdates,
Iterable<MetricUpdate<DistributionData>> distributionUpdates,
Iterable<MetricUpdate<GaugeData>> gaugeUpdates,
- Iterable<MetricUpdate<StringSetData>> stringSetUpdates) {
+ Iterable<MetricUpdate<StringSetData>> stringSetUpdates,
+ Iterable<MetricUpdate<HistogramData>> perWorkerHistogramsUpdates) {
Review Comment:
talked offline. The original comment was about the change added a metrics
named "per worker" but still based on per bundle metrics infra, now I
understand the per-worker accumulation is going to happen in runner (Dataflow
runner v2) side. If this is the case, we should name it as "histogramsUpdates"
in align with other existing metrics in Beam spec, because "per worker"
accumulation is a runner implementation, not controlled in SDK side (not part
of Beam spec).
--
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]
