Miro incident runbook

Miro Incident Runbook

Overview

Rapid incident response for Miro REST API v2 integration failures: triage, mitigation, recovery, and postmortem.

Severity Levels

Level	Definition	Response	Example
P1	Complete integration outage	< 15 min	Miro API returns 5xx on all calls
P2	Degraded service	< 1 hour	High latency, partial 429s
P3	Minor impact	< 4 hours	Webhook delays, single-board errors
P4	No user impact	Next business day	Monitoring gaps, non-critical warnings

Quick Triage (First 5 Minutes)

#!/bin/bash
# miro-triage.sh — Run this first during any Miro incident

echo "=== MIRO TRIAGE $(date -u +%H:%M:%SZ) ==="

# 1. Is Miro itself down?
echo -n "Miro Status: "
curl -sf "https://status.miro.com/api/v2/status.json" | jq -r '.status.description' 2>/dev/null || echo "STATUS PAGE UNREACHABLE"

# 2. Can we reach the API?
echo -n "API Connectivity: "
curl -s -o /dev/null -w "HTTP %{http_code} (%{time_total}s)" \
  -H "Authorization: Bearer ${MIRO_ACCESS_TOKEN}" \
  "https://api.miro.com/v2/boards?limit=1" 2>/dev/null
echo ""

# 3. What's our rate limit status?
echo "Rate Limit:"
curl -sI -H "Authorization: Bearer ${MIRO_ACCESS_TOKEN}" \
  "https://api.miro.com/v2/boards?limit=1" 2>/dev/null | \
  grep -i "x-ratelimit\|retry-after" || echo "  No rate limit headers"

# 4. Token validity
echo -n "Token: "
TOKEN_RESP=$(curl -s -H "Authorization: Bearer ${MIRO_ACCESS_TOKEN}" \
  "https://api.miro.com/v1/oauth-token" 2>/dev/null)
echo "$TOKEN_RESP" | jq -r '"scopes: \(.scopes // "INVALID"), team: \(.team.id // "N/A")"' 2>/dev/null || echo "INVALID OR EXPIRED"

# 5. Our health check
echo -n "App Health: "
curl -sf "${APP_URL:-http://localhost:3000}/health" | jq -r '.miro.status // "UNAVAILABLE"' 2>/dev/null || echo "HEALTH CHECK FAILED"

Decision Tree

Miro API returning errors?
├── YES → What status code?
│   ├── 401/403 → Token issue
│   │   ├── Token expired? → Refresh token (see below)
│   │   └── Scopes changed? → Re-authorize via OAuth flow
│   ├── 429 → Rate limited
│   │   ├── Check X-RateLimit-Remaining header
│   │   ├── Honor Retry-After header
│   │   └── Reduce request rate or enable queue
│   ├── 404 → Board/item not found
│   │   └── Verify IDs haven't changed
│   └── 500/502/503 → Miro platform issue
│       ├── Check status.miro.com
│       ├── Enable graceful degradation
│       └── Wait for Miro to resolve
└── NO → Is our integration healthy?
    ├── YES → Intermittent. Monitor for recurrence.
    └── NO → Our infrastructure issue
        ├── Check pods/containers
        ├── Check memory/CPU
        └── Check network/DNS

Immediate Actions by Error Type

401 — Token Expired

# Refresh access token
curl -s -X POST https://api.miro.com/v1/oauth/token \
  -d "grant_type=refresh_token" \
  -d "client_id=${MIRO_CLIENT_ID}" \
  -d "client_secret=${MIRO_CLIENT_SECRET}" \
  -d "refresh_token=${MIRO_REFRESH_TOKEN}" | jq

# If refresh token is also expired, user must re-authorize:
# Redirect to: https://miro.com/oauth/authorize?response_type=code&client_id=${MIRO_CLIENT_ID}&redirect_uri=${REDIRECT_URI}

403 — Insufficient Permissions

# Check what scopes the token has
curl -s -H "Authorization: Bearer ${MIRO_ACCESS_TOKEN}" \
  "https://api.miro.com/v1/oauth-token" | jq '.scopes'

# Compare with what the failed endpoint requires
# boards:read for GET endpoints
# boards:write for POST/PATCH/DELETE endpoints
# team:read / organizations:read for team/org endpoints

429 — Rate Limited

# Check current rate limit status
curl -sI -H "Authorization: Bearer ${MIRO_ACCESS_TOKEN}" \
  "https://api.miro.com/v2/boards?limit=1" | grep -i ratelimit

# Response headers:
# X-RateLimit-Limit: 100000 (credits per minute)
# X-RateLimit-Remaining: 0
# Retry-After: 30 (seconds)

# Immediate mitigation: pause all non-critical API calls
# Long-term: implement caching + webhooks (see miro-performance-tuning)

5xx — Miro Platform Issue

# 1. Confirm it's Miro-side
curl -s "https://status.miro.com/api/v2/status.json" | jq '.status'

# 2. Check for ongoing incidents
curl -s "https://status.miro.com/api/v2/incidents/unresolved.json" | \
  jq '.incidents[] | {name, status, updated_at}'

# 3. Enable graceful degradation in your app
# Feature flag: MIRO_FALLBACK_ENABLED=true
# Serve cached data, queue writes for retry when Miro recovers

Communication Templates

Internal (Slack/PagerDuty)

P[1-4] INCIDENT: Miro Integration
Status: INVESTIGATING | IDENTIFIED | MONITORING | RESOLVED
Impact: [What users experience]
Root cause: [Miro-side outage | Token expired | Rate limited | Our bug]
Action: [What we're doing now]
ETA: [Expected resolution time]
Next update: [When]

External (Status Page)

Miro Integration — Degraded Performance

We are experiencing issues with our Miro integration.
[Board sync / item creation / webhook processing] may be delayed.

Root cause: [Brief technical explanation]
Workaround: [If any — e.g., "Changes will sync when service recovers"]

Last updated: [timestamp UTC]

Post-Incident Evidence Collection

# Collect evidence for postmortem
INCIDENT_DIR="miro-incident-$(date +%Y%m%d-%H%M%S)"
mkdir -p "$INCIDENT_DIR"

# API response during incident
curl -s -H "Authorization: Bearer ${MIRO_ACCESS_TOKEN}" \
  "https://api.miro.com/v2/boards?limit=1" > "$INCIDENT_DIR/api-response.json"

# Miro status page snapshot
curl -s "https://status.miro.com/api/v2/incidents/unresolved.json" > "$INCIDENT_DIR/miro-status.json"

# Application metrics (adjust query for your Prometheus)
curl -s "http://prometheus:9090/api/v1/query_range?query=rate(miro_errors_total[5m])&start=$(date -d '2 hours ago' +%s)&end=$(date +%s)&step=60" > "$INCIDENT_DIR/error-metrics.json"

# Package (exclude tokens)
tar -czf "$INCIDENT_DIR.tar.gz" "$INCIDENT_DIR"
echo "Evidence collected: $INCIDENT_DIR.tar.gz"

Postmortem Template

## Incident: Miro [Error Type]
**Date:** YYYY-MM-DD
**Duration:** X hours Y minutes
**Severity:** P[1-4]
**Impact:** [Users affected, features impacted]

### Timeline (UTC)
- HH:MM — [First error detected by monitoring]
- HH:MM — [On-call alerted]
- HH:MM — [Root cause identified]
- HH:MM — [Mitigation applied]
- HH:MM — [Service restored]

### Root Cause
[Technical explanation — e.g., "Access token expired and refresh logic
had a bug where it used the old refresh token instead of the new one
returned in the last refresh response."]

### What Went Well
- [Monitoring detected the issue within 2 minutes]
- [Runbook was accurate and followed]

### What Went Wrong
- [Token refresh logic untested in integration tests]
- [No alerting on 401 error rate]

### Action Items
- [ ] Add integration test for token refresh flow — @owner — Due date
- [ ] Add P1 alert for miro_errors_total{error_type="auth"} > 0 — @owner — Due date
- [ ] Document token rotation procedure — @owner — Due date

Error Handling

Issue	Cause	Solution
Status page unreachable	DNS/network	Use mobile or VPN
Token refresh fails	Refresh token revoked	User must re-authorize
Rate limit persists after reset	Clock skew	Use Retry-After header, not local clock
Metrics unavailable	Prometheus down	Check application logs directly

Resources

• Miro Status Page
• Miro Developer Support
• Rate Limiting Reference

Next Steps

For data handling and compliance, see miro-data-handling.