fix: cap distinct user-agents per endpoint in GrpcUsageTracker

[AlexKehayov]

Description

GrpcUsageTracker aggregates per-interval request counts in a map keyed by (endpoint, user-agent). A known SDK's version is a client-supplied component of the user-agent (any valid SemVer value is kept as-is), so within a single logging interval a caller could send many distinct valid user-agents and grow the per-endpoint map with no upper bound, using heap in proportion to the number of distinct values.

This caps the number of distinct user-agents tracked per endpoint at MAX_AGENTS_PER_ENDPOINT (1000). Once the cap is reached, further distinct user-agents are folded into the existing UNKNOWN user-agent instead of adding new entries, so per-endpoint request totals stay exact while the number of map entries (and log lines emitted at flush) is bounded.

Behavior

• No change to normal telemetry. Real deployments see far fewer than 1000 distinct (SDK, version) pairs per endpoint per interval, so the cap is never reached and existing entries are unaffected.
• Counting for an already-tracked user-agent is unchanged (fast get path).
• The map is already swapped out every logging interval, so this only bounds growth within one interval; it does not change that reset.
• The size() check is a best-effort bound and may be exceeded slightly under concurrency, which is acceptable for a safety limit.

Testing

• Added GrpcUsageTrackerTest.testRecordInteractionCapsDistinctUserAgentsPerEndpoint: records cap+500 distinct user-agents for one endpoint and asserts the map holds cap+1 entries (the cap plus the single UNKNOWN overflow bucket), the 500 overflow interactions are folded into UNKNOWN, and the total across all keys equals the number recorded (nothing lost).
• ./gradlew :app:test --tests "...GrpcUsageTrackerTest" — 10 tests pass.

[lfdt-bot]

✅ Snyk checks have passed. No issues have been found so far.

Status	Scan Engine	Critical	High	Medium	Low	Total (0)
✅	Open Source Security	0	0	0	0	0 issues

💻 Catch issues earlier using the plugins for VS Code, JetBrains IDEs, Visual Studio, and Eclipse.

[joshmarinacci]

lgtm 👍

2 file(s) changed in commit b7a4df6

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

@@            Coverage Diff            @@
##               main   #26778   +/-   ##
=========================================
  Coverage     70.49%   70.49%           
  Complexity    11670    11670           
=========================================
  Files          2583     2583           
  Lines        108300   108305    +5     
  Branches      12110    12111    +1     
=========================================
+ Hits          76346    76351    +5     
  Misses        27977    27977           
  Partials       3977     3977           

Files with missing lines	Coverage Δ	Complexity Δ
...era/node/app/grpc/impl/usage/GrpcUsageTracker.java	98.70% <100.00%> (+0.05%)	0.00 <0.00> (ø)
...com/hedera/node/app/grpc/impl/usage/UserAgent.java	93.02% <100.00%> (+0.16%)	0.00 <0.00> (ø)
...hedera/node/app/grpc/impl/usage/UserAgentType.java	97.14% <100.00%> (+0.08%)	0.00 <0.00> (ø)

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[testlens-app]

✅ All tests passed ✅

🏷️ Commit: eb695f6
▶️ Tests: 21803 executed
⚪️ Checks: 48/48 completed

---

Learn more about TestLens at testlens.app.