[
https://issues.apache.org/jira/browse/FLINK-40292?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Weiqing Yang updated FLINK-40292:
---------------------------------
Description:
This is the first of four sub-tasks implementing FLIP-485.
FLIP:
https://cwiki.apache.org/confluence/spaces/FLINK/pages/373885706/FLIP-485+Add+UDF+Metrics
Vote thread: https://lists.apache.org/thread/symqpswsohl2s5wmtkcw0jjp1w5dot0n
Add the shared runtime helper {{UdfMetrics}} to flink-table-runtime, in the
package
org.apache.flink.table.runtime.operators.metrics. It owns metric
registration, the
sampling decision, timing, and exception counting, and it is the piece that
both the
synchronous and the asynchronous instrumentation paths use.
The helper registers two metrics under {{addGroup("udf", udfName)}} on the
executing
operator's metric group, so the full identifier is
{{<operator_name>.udf.<udf_name>.<metric>}}:
* {{udfProcessingTime}}: a DescriptiveStatisticsHistogram of per-invocation
nanoseconds.
It is sampled with the same counter-based scheme as state latency tracking
(FLINK-21736), including the "sample interval == 1" case that measures
every call.
* {{udfExceptionCount}}: a ThreadSafeSimpleCounter, incremented on every
exception and
not sampled.
The histogram is safe to update from an async callback thread. The sampling
counter is
only advanced on the task thread at dispatch.
No call site is added here. The first caller arrives in the sub-task that
adds the
configuration options and instruments synchronous scalar and table UDF calls.
Covered by a new UdfMetricsTest unit test.
was:
This is the first of four sub-tasks implementing FLIP-485: Add UDF Metrics,
which was
accepted on 2026-08-01 with 3 binding +1 and no -1.
FLIP:
https://cwiki.apache.org/confluence/spaces/FLINK/pages/373885706/FLIP-485+Add+UDF+Metrics
Vote thread: https://lists.apache.org/thread/symqpswsohl2s5wmtkcw0jjp1w5dot0n
Add the shared runtime helper {{UdfMetrics}} to flink-table-runtime, in the
package
org.apache.flink.table.runtime.operators.metrics. It owns metric
registration, the
sampling decision, timing, and exception counting, and it is the piece that
both the
synchronous and the asynchronous instrumentation paths use.
The helper registers two metrics under {{addGroup("udf", udfName)}} on the
executing
operator's metric group, so the full identifier is
{{<operator_name>.udf.<udf_name>.<metric>}}:
* {{udfProcessingTime}}: a DescriptiveStatisticsHistogram of per-invocation
nanoseconds.
It is sampled with the same counter-based scheme as state latency tracking
(FLINK-21736), including the "sample interval == 1" case that measures
every call.
* {{udfExceptionCount}}: a ThreadSafeSimpleCounter, incremented on every
exception and
not sampled.
The histogram is safe to update from an async callback thread. The sampling
counter is
only advanced on the task thread at dispatch.
No call site is added here. The first caller arrives in the sub-task that
adds the
configuration options and instruments synchronous scalar and table UDF calls.
Covered by a new UdfMetricsTest unit test.
> Add UdfMetrics helper for UDF metrics
> -------------------------------------
>
> Key: FLINK-40292
> URL: https://issues.apache.org/jira/browse/FLINK-40292
> Project: Flink
> Issue Type: Sub-task
> Components: Table SQL / Runtime
> Reporter: Weiqing Yang
> Priority: Major
>
> This is the first of four sub-tasks implementing FLIP-485.
> FLIP:
> https://cwiki.apache.org/confluence/spaces/FLINK/pages/373885706/FLIP-485+Add+UDF+Metrics
> Vote thread:
> https://lists.apache.org/thread/symqpswsohl2s5wmtkcw0jjp1w5dot0n
> Add the shared runtime helper {{UdfMetrics}} to flink-table-runtime, in the
> package
> org.apache.flink.table.runtime.operators.metrics. It owns metric
> registration, the
> sampling decision, timing, and exception counting, and it is the piece that
> both the
> synchronous and the asynchronous instrumentation paths use.
> The helper registers two metrics under {{addGroup("udf", udfName)}} on the
> executing
> operator's metric group, so the full identifier is
> {{<operator_name>.udf.<udf_name>.<metric>}}:
> * {{udfProcessingTime}}: a DescriptiveStatisticsHistogram of per-invocation
> nanoseconds.
> It is sampled with the same counter-based scheme as state latency tracking
> (FLINK-21736), including the "sample interval == 1" case that measures
> every call.
> * {{udfExceptionCount}}: a ThreadSafeSimpleCounter, incremented on every
> exception and
> not sampled.
> The histogram is safe to update from an async callback thread. The sampling
> counter is
> only advanced on the task thread at dispatch.
> No call site is added here. The first caller arrives in the sub-task that
> adds the
> configuration options and instruments synchronous scalar and table UDF
> calls.
> Covered by a new UdfMetricsTest unit test.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
