[
https://issues.apache.org/jira/browse/BEAM-6022?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Lukasz Gajowy updated BEAM-6022:
--------------------------------
Description:
I noticed that when the below (or any other metric type) is called multiple
times:
{code:java}
Metrics.counter("namespace", "name");{code}
metrics of the same name and type are created. So when such metrics are queried
by name:
{code:java}
MetricQueryResults metrics =
result
.metrics()
.queryMetrics(
MetricsFilter.builder()
.addNameFilter(MetricNameFilter.named(namespace, name))
.build());
Iterable<MetricResult<Long>> counters = metrics.getCounters();{code}
there is ambiguity which MetricResult should be chosen.
Should api users obtain one metric result here?
Should api users merge the counter results if there are multiple returned?
I noticed that all of the results contain the same value (not sure if this is
always the case). Should api users read only one of the multiple results to get
correct metric.
was:
I noticed that when the below (or any other metric type) is called:
{code:java}
Metrics.counter("namespace", "name");{code}
metrics of the same name and type are created. So when such metrics are queried
by name:
{code:java}
MetricQueryResults metrics =
result
.metrics()
.queryMetrics(
MetricsFilter.builder()
.addNameFilter(MetricNameFilter.named(namespace, name))
.build());
Iterable<MetricResult<Long>> counters = metrics.getCounters();{code}
there is ambiguity which MetricResult should be chosen.
Should api users obtain one metric result here?
Should api users merge the counter results if there are multiple returned?
I noticed that all of the results contain the same value (not sure if this is
always the case). Should api users read only one of the multiple results to get
correct metric.
> Metrics of the same name and type are created multiple times
> ------------------------------------------------------------
>
> Key: BEAM-6022
> URL: https://issues.apache.org/jira/browse/BEAM-6022
> Project: Beam
> Issue Type: Bug
> Components: sdk-java-core
> Reporter: Lukasz Gajowy
> Assignee: Kenneth Knowles
> Priority: Major
>
> I noticed that when the below (or any other metric type) is called multiple
> times:
> {code:java}
> Metrics.counter("namespace", "name");{code}
> metrics of the same name and type are created. So when such metrics are
> queried by name:
> {code:java}
> MetricQueryResults metrics =
> result
> .metrics()
> .queryMetrics(
> MetricsFilter.builder()
> .addNameFilter(MetricNameFilter.named(namespace, name))
> .build());
> Iterable<MetricResult<Long>> counters = metrics.getCounters();{code}
> there is ambiguity which MetricResult should be chosen.
> Should api users obtain one metric result here?
> Should api users merge the counter results if there are multiple returned?
> I noticed that all of the results contain the same value (not sure if this is
> always the case). Should api users read only one of the multiple results to
> get correct metric.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
