Dennis-Mircea opened a new pull request, #1078:
URL: https://github.com/apache/flink-kubernetes-operator/pull/1078
<!--
*Thank you very much for contributing to the Apache Flink Kubernetes
Operator - we are happy that you want to help us improve the project. To help
the community review your contribution in the best possible way, please go
through the checklist below, which will get the contribution into a shape in
which it can be best reviewed.*
## Contribution Checklist
- Make sure that the pull request corresponds to a [JIRA
issue](https://issues.apache.org/jira/projects/FLINK/issues). Exceptions are
made for typos in JavaDoc or documentation files, which need no JIRA issue.
- Name the pull request in the form "[FLINK-XXXX] [component] Title of the
pull request", where *FLINK-XXXX* should be replaced by the actual issue
number. Skip *component* if you are unsure about which is the best component.
Typo fixes that have no associated JIRA issue should be named following
this pattern: `[hotfix][docs] Fix typo in event time introduction` or
`[hotfix][javadocs] Expand JavaDoc for PuncuatedWatermarkGenerator`.
- Fill out the template below to describe the changes contributed by the
pull request. That will give reviewers the context they need to do the review.
- Make sure that the change passes the automated tests, i.e., `mvn clean
verify` passes. You can read more on how we use GitHub Actions for CI
[here](https://nightlies.apache.org/flink/flink-kubernetes-operator-docs-main/docs/development/guide/#cicd).
- Each pull request should address only one issue, not mix up code from
multiple issues.
- Each commit in the pull request has a meaningful commit message
(including the JIRA id)
- Once all items of the checklist are addressed, remove the above text and
this checklist, leaving only the filled out template below.
**(The sections below can be removed for hotfixes of typos)**
-->
## What is the purpose of the change
Fixes the inaccurate scaling metric computation for non-source vertices by
enabling per-second rate metrics (`numRecordsInPerSecond` /
`numRecordsOutPerSecond`) collection and consumption across the full autoscaler
pipeline, and replaces endpoint-only getRate() with spike-resilient
alternatives for gauge metrics like `LAG`.
JIRA: https://issues.apache.org/jira/browse/FLINK-39306
## Brief change log
- `FlinkMetric`: Added `NUM_RECORDS_IN_PER_SEC` and
`NUM_RECORDS_OUT_PER_SEC` enum entries to match Flink's `numRecordsInPerSecond`
/ `numRecordsOutPerSecond` task-level metrics.
- `ScalingMetric`: Added `NUM_RECORDS_IN_PER_SECOND` and
`NUM_RECORDS_OUT_PER_SECOND` scaling metric entries for storing per-second
rates in the metrics history.
- `ScalingMetricCollector`: Extended `getFilteredVertexMetricNames` to
request `NUM_RECORDS_IN_PER_SEC` and `NUM_RECORDS_OUT_PER_SEC` for non-source
vertices (previously only source-specific metrics were requested).
- `ScalingMetrics`: Extended `computeDataRateMetrics` to store
`NUM_RECORDS_IN_PER_SECOND` and `NUM_RECORDS_OUT_PER_SECOND` in the collected
scaling metrics for non-source vertices when available.
- `ScalingMetricEvaluator`:
- Introduced `getAverageWithRateFallback(perSecondMetric,
accumulatedMetric, ...)` - tries `getAverage(perSecondMetric)` first (direct
per-second rate from Flink), falls back to `getRate(accumulatedMetric)`
(endpoint-based delta from accumulated counters) when the per-second metric is
unavailable.
- Introduced `getAverageRate(metric, ...)` - computes the average of
per-interval deltas across the full metrics window, replacing the
spike-susceptible endpoint-only `getRate()` for gauge metrics like LAG.
- Replaced all `getRate(NUM_RECORDS_IN, ...)` / `getRate(NUM_RECORDS_OUT,
...)` calls in `isProcessingBacklog`, `evaluateMetrics`,
`computeEdgeOutputRatio`, and `computeEdgeDataRate` with
`getAverageWithRateFallback(NUM_RECORDS_IN_PER_SECOND, NUM_RECORDS_IN, ...)` /
`getAverageWithRateFallback(NUM_RECORDS_OUT_PER_SECOND, NUM_RECORDS_OUT, ...)`.
- Replaced `getRate(LAG, ...)` in `computeTargetDataRate` with
`getAverageRate(LAG, ...)` to avoid diluted lag rate estimation when the
metrics window is large.
- `JobAutoScalerImplTest`: Fixed flaky `testMetricReporting` by injecting a
deterministic `Clock` via `autoscaler.setClock()` to guarantee distinct
timestamps across metric collections, avoiding the timestamp collision that
caused the metrics history to stay at size 1.
## Verifying this change
This change can be verified by existing unit tests after the flaky test fix.
Additional UTs to be added to validate:
- `getAverageWithRateFallback` returns the per-second average when available
and falls back to getRate when it is not.
- `getAverageRate` computes the correct average of per-interval deltas and
is resilient to spikes diluted over a large window.
- Non-source vertices correctly collect, store, and evaluate
`NUM_RECORDS_IN_PER_SECOND` / `NUM_RECORDS_OUT_PER_SECOND`.
## Does this pull request potentially affect one of the following parts:
- Dependencies (does it add or upgrade a dependency): no
- The public API, i.e., is any changes to the `CustomResourceDescriptors`:
no
- Core observer or reconciler logic that is regularly executed: no
## Documentation
- Does this pull request introduce a new feature? no
- If yes, how is the feature documented? not applicable
--
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]
