Make adaptive routing variables runtime adjustable

[timothy-e]

Summary

Makes three adaptive routing parameters adjustable at runtime via cluster config changes:

• hybrid.score.exponent — the exponent used in the hybrid score formula (O+A+B)^N * C
• ewma.alpha — the smoothing factor for exponential moving averages (latency and in-flight requests)
• autodecay.window.ms — the time threshold after which EMA values auto-decay toward zero

When any of these configs change, the new value is propagated to all existing ServerRoutingStatsEntry instances in both the SSE and MSE stats maps. New entries created after the change also use the updated values.

Motivation

Make it easier to tune adaptive routing!

Testing

We've deployed this internally and used it to tune all three variables.

The testing we did before merging internally:

Deployed to a QA cluster.

curl -X POST localhost:9000/cluster/configs -H "Content-Type: application/json" -d '{"pinot.broker.adaptive.server.selector.ewma.alpha": "0"}'
{"status":"Updated cluster config."}

curl -X POST localhost:9000/cluster/configs -H "Content-Type: application/json" -d '{"pinot.broker.adaptive.server.selector.hybrid.score.exponent": "0"}'
{"status":"Updated cluster config."}

Harder to prove it worked, but we do see Updated EWMA alpha to 0.0 and propagated to all entries. in the logs.

curl -X POST localhost:9000/cluster/configs -H "Content-Type: application/json" -d '{"pinot.broker.adaptive.server.selector.autodecay.window.ms": "100000"}'
{"status":"Updated cluster config."}

We can see the stats stay high for a lot longer before they drop down.

[codecov-commenter]

Codecov Report

❌ Patch coverage is 84.25197% with 20 lines in your changes missing coverage. Please review.
✅ Project coverage is 64.81%. Comparing base (d469cb1) to head (c8d5cee).
⚠️ Report is 8 commits behind head on master.

Files with missing lines	Patch %	Lines
...erver/routing/stats/ServerRoutingStatsManager.java	81.92%	6 Missing and 9 partials ⚠️
...e/pinot/common/utils/ExponentialMovingAverage.java	83.33%	1 Missing and 4 partials ⚠️

Additional details and impacted files

@@             Coverage Diff              @@
##             master   #18807      +/-   ##
============================================
+ Coverage     64.78%   64.81%   +0.03%     
  Complexity     1309     1309              
============================================
  Files          3381     3386       +5     
  Lines        209967   210247     +280     
  Branches      32891    32933      +42     
============================================
+ Hits         136020   136269     +249     
- Misses        62979    63002      +23     
- Partials      10968    10976       +8     

Flag	Coverage Δ
custom-integration1	100.00% <ø> (ø)
integration	100.00% <ø> (ø)
integration1	100.00% <ø> (ø)
integration2	0.00% <ø> (ø)
java-21	64.81% <84.25%> (+0.03%)	⬆️
temurin	64.81% <84.25%> (+0.03%)	⬆️
unittests	64.81% <84.25%> (+0.03%)	⬆️
unittests1	56.99% <83.46%> (+0.03%)	⬆️
unittests2	37.25% <9.44%> (-0.01%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.