fix(seidb): report only the current migration boundary on the snapshot gauge

[blindchaser]

seidb_migration_boundary_snapshot grows without bound during a migration. Every 10 seconds MigrationMetrics recorded the current boundary as a new boundary_hex label on a synchronous OTel gauge, and the OTel Go SDK keeps every attribute set a synchronous gauge has recorded and exports all of them on each scrape. On arctic-1, each migrating node exports about 15.7k series for this metric, growing by 360 per hour, which is about 615k active series in prod Prometheus. Because every series has the value 1, no query can tell which one is the current boundary, so the metric cannot serve its purpose either.

The boundary snapshot is now an Int64ObservableGauge. registerBoundarySnapshot registers a callback, observeBoundarySnapshot, that reports only the current boundary at each collection, or complete once the version reaches targetVersion. The callback is unregistered when the router context is cancelled, so a rebuilt or closed router, including a read-only handle, stops reporting. This removes the 10-second snapshot loop and the interval parameter of NewMigrationMetrics. The metric name, value, and label stay the same.

There is no consensus, state, or wire-format impact; the change is limited to metrics. New tests in migration_metrics_test.go use a manual reader to check that only the current boundary is exported, that the label is complete at the target version, and that nothing is exported after Close. The migration, composite, and rootmulti package tests pass.

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-09-24T17:57:51.840684Z	793ab83	PR opened

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

[cursor]

PR Summary

Low Risk
Observability-only change to migration OTel metrics; no consensus, state, or migration routing behavior changes.

Overview
Fixes unbounded Prometheus cardinality on seidb_migration_boundary_snapshot: the old synchronous gauge recorded a new boundary_hex label every 10 seconds, and the OTel SDK exported every label ever seen.

MigrationMetrics now uses an Int64ObservableGauge with observeBoundarySnapshot registered via RegisterCallback, so each scrape exports only the current boundary (or complete when currentVersion reaches targetVersion). The periodic ticker loop and boundarySnapshotInterval argument on NewMigrationMetrics are removed; router builders pass NewMigrationMetrics(ctx, targetVersion) only. A small goroutine waits on context cancel/Close and unregisters the callback so rebuilt or closed routers stop reporting.

New migration_metrics_test.go checks single-label export, the complete sentinel, and no export after Close. Comment-only tweaks in composite store and migration manager describe the new gauge lifecycle.

^{Reviewed by Cursor Bugbot for commit 793ab83. Bugbot is set up for automated code reviews on this repo. Configure here.}

[github-actions]

The latest Buf updates on your PR. Results from workflow Buf / buf (pull_request).

Build	Format	Lint	Breaking	Updated (UTC)
✅ passed	✅ passed	✅ passed	✅ passed	Sep 25, 2026, 2:49 PM

[seidroid]

The PR swaps the synchronous boundary-snapshot gauge for an Int64ObservableGauge whose callback reports only the current boundary, or "complete" once the target version is reached. That fixes the unbounded series growth. The callback is unregistered when the router context is cancelled, reads its state under the existing mutex, and all three call sites are updated. The new ManualReader tests cover reporting only the current boundary, the "complete" label, and stopping after Close. I found no issues.

Findings: 0 blocking | 0 non-blocking | 0 posted inline

Blockers

• None at the file/PR level.

Non-blocking

• None at the file/PR level.

[codecov]

Codecov Report

❌ Patch coverage is 94.73684% with 1 line in your changes missing coverage. Please review.
✅ Project coverage is 66.64%. Comparing base (9903b76) to head (793ab83).
⚠️ Report is 9 commits behind head on main.

Files with missing lines	Patch %	Lines
sei-db/state_db/sc/migration/migration_metrics.go	93.75%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #4327      +/-   ##
==========================================
- Coverage   67.53%   66.64%   -0.89%     
==========================================
  Files        2181     2077     -104     
  Lines      167804   157492   -10312     
==========================================
- Hits       113325   104966    -8359     
+ Misses      54469    52516    -1953     
  Partials       10       10              

Flag	Coverage Δ
sei-db	74.50% <ø> (-0.25%)	⬇️
sei-db-state-db	?
sei-db-state-db-pr	86.30% <94.73%> (?)

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

Files with missing lines	Coverage Δ
sei-db/state_db/sc/composite/store.go	79.09% <ø> (ø)
sei-db/state_db/sc/migration/migration_manager.go	97.57% <ø> (ø)
sei-db/state_db/sc/migration/router_builder.go	76.92% <100.00%> (ø)
sei-db/state_db/sc/migration/migration_metrics.go	94.83% <93.75%> (+10.27%)	⬆️

... and 105 files with indirect coverage changes

🚀 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.

[seidroid]

Successfully created backport PR for release/v6.7:

• Backport release/v6.7: fix(seidb): report only the current migration boundary on the snapshot gauge #4348