sentinel prewarm cache

[chronark]

cold starts are pretty annoying for latency, because sentinels need multiple round trips to the db for the first request

sentinels now preload deployment and instance data to populate their cache when they boot up.

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
dashboard	Ready	Preview, Comment	Feb 23, 2026 5:04pm

1 Skipped Deployment

Project	Deployment	Actions	Updated (UTC)
engineering	Ignored	Preview	Feb 23, 2026 5:04pm

[coderabbitai]

📝 Walkthrough

Walkthrough

The change adds a cache warm-up routine to the router service that populates deployment and instances caches at startup by querying deployments with status Ready in the current environment and loading their associated instances.

Changes

Cohort / File(s)	Summary
Cache Warm-up Initialization svc/sentinel/services/router/service.go	Added database/sql import, introduced prewarm() method that queries Ready deployments and populates deploymentCache and instancesCache, invokes prewarm in New() via background goroutine with appropriate error handling, updated cache-related comments for clarity.

Estimated code review effort

🎯 3 (Moderate) | ⏱️ ~20 minutes

🚥 Pre-merge checks | ✅ 1 | ❌ 2

❌ Failed checks (2 warnings)

Check name	Status	Explanation	Resolution
Description check	⚠️ Warning	The description provides context about the problem (cold-start latency) and the solution (preloading cache), but lacks required sections from the template like issue reference, type of change, testing instructions, and checklist items.	Add the missing required sections: reference an issue number, mark the type of change, describe how to test the changes, and complete the required checklist items.
Docstring Coverage	⚠️ Warning	Docstring coverage is 0.00% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (1 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title 'sentinel prewarm cache' is concise and directly describes the main change—adding a cache prewarming mechanism to sentinels at startup.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

• 📝 Generate docstrings (stacked PR)
• 📝 Generate docstrings (commit on current branch)

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Post copyable unit tests in a comment
• Commit unit tests in branch sentinel-prewarm-cache

Warning

There were issues while running some tools. Please review the errors and either fix the tool's configuration or disable the tool if it's a critical failure.

🔧 golangci-lint (2.5.0)

Command failed

---

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[coderabbitai]

🧹 Nitpick comments (3)

svc/sentinel/services/router/service.go (3)

181-181: Consider Debug level for per-deployment logging.

logger.Info per deployment may be noisy with many deployments. Consider logger.Debug here, keeping only the summary log at Info level.

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@svc/sentinel/services/router/service.go` at line 181, Change the
per-deployment log call from logger.Info to logger.Debug to avoid noisy
Info-level output for each deployment; locate the call to
logger.Info("precaching deployment", "deployment_id", d.ID) and replace it with
logger.Debug(...) while keeping any existing summary or aggregate Info-level
logs elsewhere unchanged so only the overall summary remains at Info.

---

153-153: Consider using a cancellable context for graceful shutdown.

context.Background() means prewarm continues even if service shuts down. For a quick warm-up this is likely fine, but consider storing a cancel func to stop prewarm on Close() if deployments are numerous.

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@svc/sentinel/services/router/service.go` at line 153, The prewarm is started
with context.Background() so it won't stop on shutdown; change it to use a
cancellable context stored on the service (e.g. create ctx, cancel :=
context.WithCancel(ctx) during service start/Init and assign to s.prewarmCtx and
s.prewarmCancel) and call s.prewarm(s.prewarmCtx) instead of
s.prewarm(context.Background()); ensure Close() calls the stored cancel
(s.prewarmCancel()) to stop the goroutine and clean up.

---

157-169: Consider adding a timeout to prewarm.

No timeout on DB operations. If the database is slow or unresponsive, prewarm could hang indefinitely. Consider wrapping with a deadline context.

♻️ Suggested fix

 func (s *service) prewarm(ctx context.Context) {
 	logger.Info("prewarming cache")
+
+	ctx, cancel := context.WithTimeout(ctx, 30*time.Second)
+	defer cancel()

 	deployments, err := db.Query.ListDeploymentsByEnvironmentIdAndStatus(ctx, s.db.RO(), db.ListDeploymentsByEnvironmentIdAndStatusParams{

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@svc/sentinel/services/router/service.go` around lines 157 - 169, The prewarm
function lacks a deadline for the database call which can block forever; wrap
the DB query in a context with timeout (e.g., ctxTimeout, cancel :=
context.WithTimeout(ctx, <reasonable-duration>) and defer cancel()) and pass
ctxTimeout to db.Query.ListDeploymentsByEnvironmentIdAndStatus (instead of ctx)
so the call on s.db.RO() will be canceled if it exceeds the timeout; consider
making the timeout configurable on the service and handle
context.DeadlineExceeded similarly to other errors.

ℹ️ Review info

Configuration used: Repository UI

Review profile: CHILL

Plan: Pro

📥 Commits

Reviewing files that changed from the base of the PR and between 71105b2 and f2b5184.

📒 Files selected for processing (1)

• svc/sentinel/services/router/service.go

🤖 Prompt for all review comments with AI agents

Verify each finding against the current code and only fix it if needed.

Nitpick comments:
In `@svc/sentinel/services/router/service.go`:
- Line 181: Change the per-deployment log call from logger.Info to logger.Debug
to avoid noisy Info-level output for each deployment; locate the call to
logger.Info("precaching deployment", "deployment_id", d.ID) and replace it with
logger.Debug(...) while keeping any existing summary or aggregate Info-level
logs elsewhere unchanged so only the overall summary remains at Info.
- Line 153: The prewarm is started with context.Background() so it won't stop on
shutdown; change it to use a cancellable context stored on the service (e.g.
create ctx, cancel := context.WithCancel(ctx) during service start/Init and
assign to s.prewarmCtx and s.prewarmCancel) and call s.prewarm(s.prewarmCtx)
instead of s.prewarm(context.Background()); ensure Close() calls the stored
cancel (s.prewarmCancel()) to stop the goroutine and clean up.
- Around line 157-169: The prewarm function lacks a deadline for the database
call which can block forever; wrap the DB query in a context with timeout (e.g.,
ctxTimeout, cancel := context.WithTimeout(ctx, <reasonable-duration>) and defer
cancel()) and pass ctxTimeout to
db.Query.ListDeploymentsByEnvironmentIdAndStatus (instead of ctx) so the call on
s.db.RO() will be canceled if it exceeds the timeout; consider making the
timeout configurable on the service and handle context.DeadlineExceeded
similarly to other errors.