What is a canary deploy

A canary deploy is the practice of exposing a new version of software to a small percentage of users before doing a full rollout. If something goes wrong, you revert instantly — without redeploying.

With CanaryGate, you do canary deploys at the feature level, not the infrastructure level.

Setting up the canary

1. Create a rollout flag

// In the dashboard: New Flag → Type: Rollout
// key: "v2-engine"

2. Initialize with userId

const client = new CanaryGate(process.env.CANARYGATE_KEY!, {
  userId: currentUser.id // Stable ID — the same user always sees the same result
})
await client.init()

3. Ramp up gradually

Adjust the percentage in the dashboard. The change reaches clients in real time.

4. Evaluate in code

const useV2 = client.getFlag('v2-engine')?.enabled ?? false
 
if (useV2) {
  return await processWithV2Engine(data)
} else {
  return await processWithV1Engine(data)
}

Instant rollback

If you detect a problem, set the percentage back to 0% in the dashboard — no redeployment, no downtime.

With auto-rollout (Pro plan)

Configure automatic percentage increases with a schedule:

10:00 → 5%
14:00 → 25%
next day 09:00 → 50%
48h later → 100%

CanaryGate executes each step automatically. You can still pause or revert manually at any time.

Cleanup

Once the feature is at 100% and stable, remove the flag from the code and delete it in the dashboard to keep the project tidy.

// Before (with flag):
const isNewEngine = client.getFlag('v2-engine')?.enabled ?? false
if (isNewEngine) { ... }
 
// After (100% rollout, flag removed):
// Permanent code without conditional