Enable dynamic toggling of adaptive routing stats metrics export#18135
Draft
timothy-e wants to merge 2 commits intoapache:masterfrom
Draft
Enable dynamic toggling of adaptive routing stats metrics export#18135timothy-e wants to merge 2 commits intoapache:masterfrom
timothy-e wants to merge 2 commits intoapache:masterfrom
Conversation
cc stripe-private-oss-forks/pinot-reviewers r? When multistage adaptive server selector stats collection is enabled (pinot.broker.adaptive.server.selector.enable.stats.collection=true), the broker tracks per-server routing stats (in-flight request count, latency EMA, hybrid score) in memory but has no way to observe them externally without hitting the broker's debug API. This makes it difficult to monitor the health and behavior of adaptive routing in production. This PR adds `setValueOfGaugeWithTag(gauge, tag, value)` to `AbstractMetrics`, which composes the metric name as `gaugeName.tag` without going through `getTableName()`. This ensures the server instance ID is always present in the metric name regardless of the `enableTableLevelMetrics` broker config. Also adds `BrokerMetrics.getTagForServer(serverInstanceId)` returning "server.<instanceId>", following the same pattern as `getTagForPreferredPool`. The resulting Pinot metric names are of the form: `pinot.broker.adaptiveServerLatencyEma.server.Server_pinotdb1_8098` **PluginConfig.yaml + statsd reporter** Following https://trailhead.corp.stripe.com/docs/stream-analytics-internal/pinot/pinot-observability/pinot-metrics-guide, adds a labeled metric pattern that extracts the server instance ID as a server tag, so the metrics arrive in Veneur/Prometheus as: - Name: pinot_broker_adaptive_server_latency_ema - Tags: server=Server_pinotdb1_8098 (plus standard host tags) [STREAMANALYTICS-4390](https://jira.corp.stripe.com/browse/STREAMANALYTICS-4390) Unit tests to verify that we: * don't export when adaptive routing stats are disabled * don't export when adaptive routing stats metric export is disabled * transform the metric name properly. [Deployed to QA rad-canary](https://amp.qa.corp.stripe.com/deploy/qa-deploy1.pdx.deploy.stripe.net%2Fdeploy_qBdrdOKZQ7iya8JplJbdaw), and [metrics appeared in grafana](https://g-8916660cfe.grafana-workspace.us-west-2.amazonaws.com/explore?schemaVersion=1&panes=%7B%223yy%22:%7B%22datasource%22:%22zb219lV4k%22,%22queries%22:%5B%7B%22refId%22:%22B%22,%22expr%22:%22%7B__name__%3D~%5C%22pinot_broker_adaptive_server_%28latency_ema%7Cnum_in_flight_requests%7Chybrid_score%29%5C%22,%5Cn%20%20pinot_cluster%3D%5C%22rad-canary%5C%22,host%3D%5C%22qa-pinotdbbroker--06621e6804f2c92db.northwest.stripe.io%5C%22%7D%22,%22range%22:true,%22instant%22:true,%22datasource%22:%7B%22type%22:%22prometheus%22,%22uid%22:%22zb219lV4k%22%7D,%22editorMode%22:%22code%22,%22legendFormat%22:%22__auto%22%7D%5D,%22range%22:%7B%22from%22:%221775151966533%22,%22to%22:%221775152978766%22%7D%7D%7D&orgId=1) as `{__name__=~"pinot_broker_adaptive_server_(latency_ema|num_in_flight_requests|hybrid_score)", pinot_cluster="rad-canary"}`. Stripe-Original-Repo: stripe-private-oss-forks/pinot Stripe-Monotonic-Timestamp: v2/2026-04-07T17:09:47Z/0 Stripe-Original-PR: https://git.corp.stripe.com/stripe-private-oss-forks/pinot/pull/581 (cherry picked from commit 8d84c8980f2122e5290b15d3889282eee8f5286f)
Committed-By-Agent: claude cc stripe-private-oss-forks/pinot-reviewers r? https://git.corp.stripe.com/stripe-private-oss-forks/pinot/pull/581 added metrics for adaptive routing stats. We may want to toggle whether or not stats are exported as metrics without needing to launch a rolling restart. ServerRoutingStatsManager now implements PinotClusterConfigChangeListener. The periodic export task is always scheduled when stats collection is enabled; the enable.stats.metric.export flag is checked inside the task and can be updated at runtime via the Helix cluster config (PUT /cluster/configs) without a broker restart. We also allow updating the metrics export frequency at runtime. [STREAMANALYTICS-4390](https://jira.corp.stripe.com/browse/STREAMANALYTICS-4390) Deployed to [rad-canary QA](https://amp.qa.corp.stripe.com/deploy/qa-deploy1.pdx.deploy.stripe.net%2Fdeploy_r2WUckQcQ467TfeJvyp4zw). `ssh`ed onto a rad-canary controller and ran ``` curl -X POST localhost:9000/cluster/configs -H "Content-Type: application/json" -d '{"pinot.broker.adaptive.server.selector.enable.stats.metric.export": "false"}' curl -X POST localhost:9000/cluster/configs -H "Content-Type: application/json" -d '{"pinot.broker.adaptive.server.selector.enable.stats.metric.export": "true"}' curl -X POST localhost:9000/cluster/configs -H "Content-Type: application/json" -d '{"pinot.broker.adaptive.server.selector.enable.stats.metric.export": "false"}' ``` I see that the metrics stopped / resumed as expected in [grafana](https://g-8916660cfe.grafana-workspace.us-west-2.amazonaws.com/explore?schemaVersion=1&panes=%7B%223yy%22:%7B%22datasource%22:%22zb219lV4k%22,%22queries%22:%5B%7B%22refId%22:%22B%22,%22expr%22:%22count%20by%20%28host%29%20%28%7B__name__%3D~%5C%22pinot_broker_adaptive_server_latency_ema%5C%22,%20pinot_cluster%3D%5C%22rad-canary%5C%22%7D%29%22,%22range%22:true,%22instant%22:true,%22datasource%22:%7B%22type%22:%22prometheus%22,%22uid%22:%22zb219lV4k%22%7D,%22editorMode%22:%22code%22,%22legendFormat%22:%22__auto%22%7D%5D,%22range%22:%7B%22from%22:%221775159834995%22,%22to%22:%221775160734995%22%7D%7D%7D&orgId=1) <img width="1477" alt="Screenshot 2026-04-02 at 4 17 58 pm" src="https://git.corp.stripe.com/user-attachments/assets/dca06b6d-8b6f-4797-a079-95a1f21d7de7" /> Stripe-Original-Repo: stripe-private-oss-forks/pinot Stripe-Monotonic-Timestamp: v2/2026-04-07T21:12:57Z/0 Stripe-Original-PR: https://git.corp.stripe.com/stripe-private-oss-forks/pinot/pull/582 (cherry picked from commit 3abc8faad5987ed824b6b7a11afa5693dbd2c831)
timothy-e
force-pushed
the
timothye-enable-adaptive-routing-metrics-dynamic-toggle
branch
from
d34c964 to
2b4eec6
Compare
Contributor
Author
|
Based on #18134, but can't do cross-repo stacked PRs in github, so leaving this as a draft with conflicts for now. |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
#18134 adds metrics for adaptive routing stats.
We may want to toggle whether or not stats are exported as metrics without needing to rolling restart the brokers.
ServerRoutingStatsManager now implements PinotClusterConfigChangeListener. The periodic export task is always scheduled when stats collection is enabled; the enable.stats.metric.export flag is checked inside the task and can be updated at runtime via the Helix cluster config (PUT /cluster/configs) without a broker restart.
We also allow updating the metrics export frequency at runtime.
Testing
Deployed to an internal Stripe cluster.
sshed onto one of the controllers and ranSaw the metrics start/stop as expected.