Monitor Error Rates During Feature Flag Rollouts
Ship safely — when you increase a feature flag's rollout, watch error rates for that cohort and alert if they spike so you can roll back fast.
- 1
Trigger on Rollout Changes
Add a
Webhooknode fired when a flag's rollout percentage increases. - 2
Sample Error Rates
Add
Schedule TriggerandHTTP Requestnodes checking the exposed cohort's error rate over the next period. - 3
Compare to Baseline
Add a
Codenode comparing the cohort's errors to the control group. - 4
Alert on Spikes
Add an
IFandSlacknode warning the team if errors rise significantly so they can roll back. - 5
Activate and Test
Activate the workflow and simulate a rollout. Confirm monitoring and alerting work.
Frequently asked questions
Can it auto-roll back?
For severe spikes you can trigger a flag revert, but confirm with an on-call engineer for anything ambiguous.
How long should I monitor?
Watch for a meaningful window after each increase — long enough to catch delayed or load-dependent errors.