Back to all use cases

ConsultingCross-Industry

One error-handling pattern, so a failed recipe is a notification and not a surprise

Integration Team

A typical Workato consulting engagement for an integration team whose recipes fail without anyone being told: a review of Handle errors steps, retry rules by step type, alerting and API recipe timeouts, ending in one error-handling pattern.

Workato consulting

Overview

This is a typical engagement for an integration team that has built a few hundred Workato recipes over several years, each with its own idea of error handling. Failures reach the business through a downstream team, not through an alert. AceMQ engineers, with a partner subject-matter expert where needed, review the recipes and write the pattern.

Challenge

Workato's Handle errors step works like try and catch, with a Monitor block and an On error block. Retries are off by default, can be set to at most three with a wait of one to ten seconds, and can be conditional. In an API recipe, a request timeout ends the job without running the On error steps. The on-error datapills include error_type_id, which is stable, and inner_message, which is unprocessed and can change format without notice, so it should not drive logic.

Environment

A Workato workspace with workflow recipes and API recipes, callable recipes, connections to SaaS applications and an on-prem system, and job history that nobody reviews.

Approach

AceMQ engineers sample recipes by type and by system, read their Handle errors steps and job error history, and classify errors as user errors or platform-side errors. They define retry rules per step type, including which writes are unsafe to retry, and design alerting on error_type_id with routing to an owner.

Solution

  1. Recipe sample by type and system with Handle errors coverage and gaps
  2. Error history classified into user errors and platform-side errors
  3. Retry rules per step type, with the writes that must not be retried
  4. Alerting on error_type_id and routing to an owner, with inner_message kept out of logic
  5. API recipe timeout handling: client timeouts, concurrency and monitoring outside the recipe
  6. A callable recipe for shared error handling, and a rollout order starting with the riskiest recipes

Outcome

The engagement delivers a written review with the error-handling pattern, retry rules, alert routing, the list of recipes to change in order and a rehearsal plan in a test environment. The customer's team applies it, or AceMQ can apply it alongside them.

Technologies

WorkatoREST APIs

Need Workato Architecture Guidance?

AceMQ's senior Workato engineers have handled this exact type of engagement before. Whether you need architectural guidance, hands-on remediation, or an ongoing managed partnership, we're ready to help.