Apify Production Checklist — Worked Examples
Concrete examples of running the checklist against a real Actor, plus the console output each monitoring helper produces.
Example 1: First production deploy of a scraper
You have username/product-scraper passing locally. Run the checklist:
- Read
.actor/actor.jsonand confirmname,title,descriptionare set. - Confirm
Dockerfilepinsapify/actor-node:20(notlatest). apify push, thenapify builds lsto confirm the build succeeded.- Smoke-test on-platform with a tiny input:
apify actors call username/product-scraper \
--input='{"startUrls":[{"url":"https://target.com"}],"maxItems":10}'- Create the daily schedule and completion webhook (see implementation.md Steps 2–3).
Example 2: Health check output
Running checkActorHealth('username/product-scraper', 24) prints:
Actor: username/product-scraper
Last 24h: 3 runs, 66.7% success
Failed: 1, Timed out: 0
Total cost: $0.4213
ALERT: Failed runs detected!A success rate below 100% with ALERT: present is your cue to open the failed
run's log in Apify Console and cross-reference the Error Handling table in SKILL.md.
Example 3: Cost guard aborting a runaway run
Calling runWithCostGuard(actorId, input, 5.0) on a job that blows past budget:
Cost guard: $5.0018 exceeds $5. Aborting.The run is aborted mid-flight, waitForFinish() returns the aborted run object,
and the poller is cleared — no orphaned interval.
Example 4: Rolling back a bad build
Build 42 shipped a regression. Roll back to build 41 without redeploying code:
apify builds ls
curl -X POST \
-H "Authorization: Bearer $APIFY_TOKEN" \
"https://api.apify.com/v2/acts/username~product-scraper?build=41"The Actor's default build pointer moves back to 41; the next scheduled run uses it.