cloudthinker-ai Canary Deployment Skill
Execute canary deployment of {{ service_name }} {{ version }} against baseline {{ baseline_version }}.
Workflow
Step 1 — Canary Configuration
CANARY SETUP
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Service: {{ service_name }}
Canary version: {{ version }}
Baseline version: {{ baseline_version }}
Traffic ramp schedule:
Stage 1: 1% canary (10 min bake)
Stage 2: 5% canary (15 min bake)
Stage 3: 25% canary (30 min bake)
Stage 4: 50% canary (30 min bake)
Stage 5: 100% canary (promotion)
Rollback trigger (any stage):
- Error rate: canary > baseline + 1%
- P95 latency: canary > baseline * 1.5
- P99 latency: canary > baseline * 2.0
- Custom metric threshold breach
Step 2 — Pre-Canary Checks
PRE-CANARY
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
[ ] Baseline metrics captured:
- Error rate: ___%
- P50 latency: ___ms
- P95 latency: ___ms
- P99 latency: ___ms
- Throughput: ___ rps
[ ] Canary version deployed (0% traffic)
[ ] Canary pods/instances healthy
[ ] Canary version verified (version endpoint)
[ ] Monitoring dashboard with canary vs baseline comparison ready
[ ] Alerting configured for canary metrics
[ ] Traffic splitting mechanism configured (service mesh / LB / ingress)
Step 3 — Stage 1: 1% Traffic (Smoke Test)
STAGE 1 — 1% TRAFFIC (10 min bake)
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
[ ] Route 1% traffic to canary
[ ] Start time: ___
Metrics gate (at T+10min):
[ ] Error rate: canary ___% vs baseline ___% — PASS/FAIL
[ ] P95 latency: canary ___ms vs baseline ___ms — PASS/FAIL
[ ] P99 latency: canary ___ms vs baseline ___ms — PASS/FAIL
[ ] No new error types in canary logs
[ ] No panics, OOMs, or crashes
Decision: PROMOTE / ROLLBACK
Step 4 — Stage 2: 5% Traffic
STAGE 2 — 5% TRAFFIC (15 min bake)
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
[ ] Route 5% traffic to canary
[ ] Start time: ___
Metrics gate (at T+15min):
[ ] Error rate: canary ___% vs baseline ___% — PASS/FAIL
[ ] P95 latency: canary ___ms vs baseline ___ms — PASS/FAIL
[ ] P99 latency: canary ___ms vs baseline ___ms — PASS/FAIL
[ ] Memory usage stable (no leak pattern)
[ ] CPU usage proportional to traffic share
Decision: PROMOTE / ROLLBACK
Step 5 — Stage 3: 25% Traffic
STAGE 3 — 25% TRAFFIC (30 min bake)
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
[ ] Route 25% traffic to canary
[ ] Start time: ___
Metrics gate (at T+30min):
[ ] Error rate: canary ___% vs baseline ___% — PASS/FAIL
[ ] P95 latency: canary ___ms vs baseline ___ms — PASS/FAIL
[ ] P99 latency: canary ___ms vs baseline ___ms — PASS/FAIL
[ ] Business metrics normal (conversion, success rates)
[ ] Dependency health unaffected
[ ] No customer-reported issues
Decision: PROMOTE / ROLLBACK
Step 6 — Stage 4: 50% Traffic
STAGE 4 — 50% TRAFFIC (30 min bake)
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
[ ] Route 50% traffic to canary
[ ] Start time: ___
Metrics gate (at T+30min):
[ ] Error rate: canary ___% vs baseline ___% — PASS/FAIL
[ ] P95 latency: canary ___ms vs baseline ___ms — PASS/FAIL
[ ] P99 latency: canary ___ms vs baseline ___ms — PASS/FAIL
[ ] Autoscaling behaving correctly under load
[ ] All downstream dependencies stable
[ ] No data consistency issues
Decision: PROMOTE TO 100% / ROLLBACK
Step 7 — Stage 5: Full Promotion
PROMOTION — 100% TRAFFIC
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
[ ] Route 100% traffic to {{ version }}
[ ] Scale down baseline ({{ baseline_version }}) instances
[ ] Update deployment to {{ version }} as new baseline
[ ] Monitor for 30 minutes post-promotion
[ ] Confirm all metrics stable at full traffic
[ ] Deployment marked as SUCCESSFUL
[ ] Stakeholders notified
Rollback Procedure
ROLLBACK (any stage)
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
[ ] Route 100% traffic back to baseline ({{ baseline_version }})
[ ] Scale down canary instances
[ ] Verify baseline metrics return to normal
[ ] Create issue to investigate canary failure
[ ] Document: stage failed, metrics at failure, root cause hypothesis
Counter-Rationalizations
| Shortcut | Counter | Why |
|---|---|---|
| "We can skip some steps for this case" | Adapt the workflow steps, don't skip them | Skipped steps are where incidents and oversights originate |
| "The user seems to already know what to do" | Complete all workflow phases with the user | The workflow catches blind spots that experience alone misses |
| "This is a minor case, full process is overkill" | Scale the process down, don't turn it off | Minor cases become major when unstructured; the process scales, not disappears |
| "I'll fill in the details later" | Complete each section before moving on | Deferred details are forgotten; real-time capture is more accurate |
| "The template output isn't necessary" | Always produce the structured output format | Structured output enables comparison, audit trails, and handoff to other teams |
Output Format
Produce a canary deployment report with:
- Canary configuration (versions, ramp schedule, thresholds)
- Per-stage metrics comparison (canary vs baseline)
- Promotion/rollback decision at each stage with reasoning
- Final status (PROMOTED / ROLLED BACK at stage X)
- Metrics summary chart showing canary vs baseline over time