Imported from jeremylongshore/tons-of-skills-marketplace (
plugins/saas-packs/algolia-pack/skills/algolia-prod-checklist/SKILL.md). Install upstream withnpx skills add jeremylongshore/tons-of-skills-marketplace --skill algolia-prod-checklist. Copyright stays with the author (MIT).
Algolia Production Checklist
Overview
Complete checklist for deploying Algolia search to production. Covers index configuration, API key security, replica setup, monitoring, and rollback procedures.
Prerequisites
- A named production index, deployment owner, rollback decision maker, and change window.
- Read-only verification credentials plus separate, least-privileged deployment credentials.
- Current monitoring and alert destinations tested before the release begins.
Instructions
Run the checklist in order, record each result, and stop the deployment on any failed security, relevance, or availability requirement. Use the verification script only after every prerequisite configuration item is confirmed.
Examples
The checklist and pre-deploy script are the release example: run them against the intended production index and attach the results to the deployment record before enabling the new search experience.
Pre-Production Checklist
Index Configuration
-
searchableAttributesordered by priority (first = highest) -
attributesForFacetingset for all filterable attributes -
customRankingconfigured (business metrics as tie-breakers) -
unretrievableAttributesset for fields that should be searchable but not returned -
attributesToRetrievelimited to fields needed by the UI -
typoTolerancetested (default: enabled, min 4 chars for 1 typo, min 8 for 2) -
removeStopWordsconfigured for your language(s) -
distinctset if deduplication needed (e.g., one result per product group)
// Verify production settings
const settings = await client.getSettings({ indexName: 'products' });
console.log(JSON.stringify(settings, null, 2));
API Key Security
- Admin key in backend env vars only (never frontend)
- Search-Only key used in frontend with
referersrestriction -
maxQueriesPerIPPerHourset on all public keys -
maxHitsPerQuerylimited on search keys - Secured API keys used for multi-tenant data isolation
- Keys restricted to specific
indexeswhere possible
Replicas (Alternate Sorting)
// Replicas give users alternate sort orders
await client.setSettings({
indexName: 'products',
indexSettings: {
// Standard replicas: share parent's data, use their own relevance settings
replicas: [
'products_price_asc', // Sort by price ascending
'products_price_desc', // Sort by price descending
'products_newest', // Sort by newest first
],
},
});
// Configure each replica's ranking
await client.setSettings({
indexName: 'products_price_asc',
indexSettings: {
ranking: [
'asc(price)', // Primary: price ascending
'typo', 'geo', 'words', 'filters', 'proximity', 'attribute', 'exact', 'custom',
],
},
});
Monitoring
- Health check endpoint tests Algolia connectivity
- Alert on error rate > 1% over 5 minutes
- Alert on P95 latency > 200ms (Algolia is typically < 50ms)
- Dashboard shows queries/sec, latency, error rate
- status.algolia.com RSS/webhook configured
// Health check endpoint
async function algoliaHealthCheck() {
const start = Date.now();
try {
const { items } = await client.listIndices();
const latencyMs = Date.now() - start;
return {
status: 'healthy',
latencyMs,
indexCount: items.length,
totalRecords: items.reduce((sum, i) => sum + (i.entries || 0), 0),
};
} catch (error) {
return { status: 'unhealthy', error: String(error), latencyMs: Date.now() - start };
}
}
Graceful Degradation
// If Algolia is down, fall back to database search
async function searchWithFallback(query: string) {
try {
const { hits } = await client.searchSingleIndex({
indexName: 'products',
searchParams: { query, hitsPerPage: 20 },
});
return { source: 'algolia', results: hits };
} catch (error) {
console.error('Algolia unavailable, falling back to DB', error);
const dbResults = await db.products.find({
name: { $regex: query, $options: 'i' },
}).limit(20);
return { source: 'database', results: dbResults };
}
}
Pre-Deploy Verification Script
#!/bin/bash
echo "=== Algolia Production Pre-Flight ==="
# 1. Verify connectivity
HTTP_CODE=$(curl -s -o /dev/null -w "%{http_code}" \
"https://${ALGOLIA_APP_ID}-dsn.algolia.net/1/indexes" \
-H "X-Algolia-Application-Id: ${ALGOLIA_APP_ID}" \
-H "X-Algolia-API-Key: ${ALGOLIA_ADMIN_KEY}")
echo "API connectivity: HTTP $HTTP_CODE"
[ "$HTTP_CODE" != "200" ] && echo "FAIL: Cannot reach Algolia" && exit 1
# 2. Check Algolia service status
STATUS=$(curl -s https://status.algolia.com/api/v2/status.json | jq -r '.status.indicator')
echo "Algolia status: $STATUS"
[ "$STATUS" != "none" ] && echo "WARNING: Algolia reporting issues"
# 3. Verify index has data
RECORDS=$(curl -s "https://${ALGOLIA_APP_ID}-dsn.algolia.net/1/indexes/products" \
-H "X-Algolia-Application-Id: ${ALGOLIA_APP_ID}" \
-H "X-Algolia-API-Key: ${ALGOLIA_ADMIN_KEY}" | jq '.entries')
echo "Products index: $RECORDS records"
[ "$RECORDS" -lt 1 ] && echo "FAIL: Index is empty" && exit 1
echo ""
echo "All checks passed. Ready to deploy."
Output
The release record contains completed configuration, security, monitoring, degradation, and verification checks, with failed items converted into explicit blockers or rollback actions. It does not authorize deployment when a required check is unknown.
Error Handling
| Alert | Condition | Severity | Action |
|---|---|---|---|
| Search errors | 5xx or 403 errors > 5/min | P1 | Check API keys, Algolia status |
| High latency | P95 > 200ms for 5+ min | P2 | Check index size, network |
| Rate limited | 429 errors > 10/min | P2 | Reduce request rate, check key limits |
| Index stale | Last updated > 1 hour ago | P3 | Check sync pipeline |
Resources
Next Steps
For version upgrades, see algolia-upgrade-migration.