Ideogram Incident Runbook

Ideogram Incident Runbook

Overview

Rapid incident response for Ideogram API outages, auth failures, rate limiting, and degraded generation quality. Covers triage, immediate remediation, fallback activation, and postmortem process.

Severity Levels

Quick Triage (Run These First)

set -euo pipefail

echo "=== IDEOGRAM TRIAGE ==="

# 1. Test API connectivity and auth
echo -n "API status: "
curl -s -o /dev/null -w "%{http_code}" \
  -X POST https://api.ideogram.ai/generate \
  -H "Api-Key: $IDEOGRAM_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"image_request":{"prompt":"triage test","model":"V_2_TURBO","magic_prompt_option":"OFF"}}'
echo ""

# 2. Test V3 endpoint
echo -n "V3 status: "
curl -s -o /dev/null -w "%{http_code}" \
  -X POST https://api.ideogram.ai/v1/ideogram-v3/generate \
  -H "Api-Key: $IDEOGRAM_API_KEY" \
  -F "prompt=triage test" -F "rendering_speed=FLASH"
echo ""

# 3. Check DNS resolution
echo -n "DNS: "
nslookup api.ideogram.ai 2>/dev/null | grep -A1 "Name:" | tail -1 || echo "lookup failed"

# 4. Measure latency
echo -n "Latency: "
curl -s -o /dev/null -w "%{time_total}s" \
  -X POST https://api.ideogram.ai/generate \
  -H "Api-Key: $IDEOGRAM_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"image_request":{"prompt":"latency test","model":"V_2_TURBO","magic_prompt_option":"OFF"}}'
echo ""

Decision Tree

Is api.ideogram.ai returning errors?
├─ YES: What status code?
│   ├─ 401 → Key revoked or misconfigured. See "Auth Failure" below.
│   ├─ 402 → Credits exhausted. Top up immediately.
│   ├─ 422 → Safety filter. Prompt issue, not outage.
│   ├─ 429 → Rate limited. Reduce concurrency.
│   ├─ 500/503 → Ideogram outage. Enable fallback.
│   └─ Timeout → Network or Ideogram performance issue.
├─ NO: Are images generating but quality is bad?
│   ├─ YES → Check model version, style params, magic_prompt setting.
│   └─ NO → Check image download (URLs may have expired).
└─ Not sure: Run triage script above.

Immediate Actions

401 -- Authentication Failure

set -euo pipefail
# Verify key is set
echo "Key present: ${IDEOGRAM_API_KEY:+YES}${IDEOGRAM_API_KEY:-NO}"
echo "Key length: ${#IDEOGRAM_API_KEY}"

# If key was rotated, update everywhere:
# 1. Ideogram dashboard: create new key
# 2. Update secret manager / env vars
# 3. Restart affected services

# Kubernetes
kubectl create secret generic ideogram-secrets \
  --from-literal=api-key="$NEW_KEY" \
  --dry-run=client -o yaml | kubectl apply -f -
kubectl rollout restart deployment/ideogram-service

429 -- Sustained Rate Limiting

set -euo pipefail
# Reduce concurrency immediately
kubectl set env deployment/ideogram-service IDEOGRAM_CONCURRENCY=3

# If sustained, contact Ideogram for limit increase
# partnership@ideogram.ai

500/503 -- Ideogram Outage

set -euo pipefail
# Enable fallback mode (return placeholder images)
kubectl set env deployment/ideogram-service IDEOGRAM_FALLBACK=true
kubectl rollout restart deployment/ideogram-service

# Monitor for resolution, then disable fallback
# kubectl set env deployment/ideogram-service IDEOGRAM_FALLBACK=false

402 -- Credits Exhausted

1. Log into ideogram.ai > Settings > API Beta
2. Check current balance
3. Increase auto top-up amount
4. Or manually add credits
5. Verify generation works again

Fallback Implementation

const FALLBACK_ENABLED = process.env.IDEOGRAM_FALLBACK === "true";

async function generateWithFallback(prompt: string, options: any = {}) {
  if (FALLBACK_ENABLED) {
    return {
      data: [{
        url: `https://placehold.co/1024x1024/333/fff?text=${encodeURIComponent("Image unavailable")}`,
        seed: 0,
        resolution: "1024x1024",
        is_image_safe: true,
        fallback: true,
      }],
    };
  }

  try {
    return await generateImage(prompt, options);
  } catch (err: any) {
    if (err.status >= 500) {
      console.error("Ideogram 5xx -- serving fallback");
      return generateWithFallback(prompt, options);
    }
    throw err;
  }
}

Communication Templates

Internal (Slack)

P[X] INCIDENT: Ideogram Integration
Status: INVESTIGATING / MITIGATED / RESOLVED
Impact: [e.g., Image generation unavailable for users]
Cause: [e.g., API returning 500, or key revoked]
Action: [e.g., Fallback enabled, monitoring for resolution]
Next update: [time]
Owner: @[name]

Postmortem Template

## Incident: Ideogram [Type]
**Date:** YYYY-MM-DD | **Duration:** Xh Ym | **Severity:** P[1-4]

### Summary
[1-2 sentences]

### Timeline
- HH:MM - First alert triggered
- HH:MM - Triage started
- HH:MM - Fallback enabled
- HH:MM - Root cause identified
- HH:MM - Resolved

### Root Cause
[Technical explanation]

### Action Items
- [ ] [Fix] - Owner - Due date
- [ ] [Prevention] - Owner - Due date

Error Handling

Output

• Incident identified and categorized by severity
• Immediate remediation applied
• Fallback activated if needed
• Stakeholders notified with template
• Evidence collected for postmortem

Prerequisites

• An incident owner, a synthetic prompt with no protected content, approved support path, and redaction policy for prompt, asset, and output fields.

Instructions

1. Open a timestamped incident, classify impact, and reproduce once with the synthetic prompt.
2. Isolate identity, endpoint, content-policy, quota, generation, storage, or destination failure using status and opaque correlation IDs only.
3. Freeze nonessential rollout changes, apply one reversible remediation, and monitor the synthetic probe and aggregate metrics.
4. Stop if a change expands source, destination, or output-retention scope; close only after rollback decision and owner review.

Examples

P2; integration=staging-generator; error=429; action=bounded-backoff; fixture=fictional; output_retention=none; rollback=not-needed is a safe incident summary.

Resources

• Ideogram API Overview
• Enterprise support: partnership@ideogram.ai

Next Steps

For data handling patterns, see ideogram-data-handling.