Camunda 8 operations manual: workers, Operate and Tasklist

Use a connector for a supported integration and a job worker for custom service logic. A worker activates jobs of a configured type, reads required variables and reports completion or failure through the supported client API. Match the BPMN job type and worker configuration.

A worker can encounter retries or uncertain network outcomes. Design external side effects to be idempotent and record stable business identifiers. Apply appropriate timeouts and least-privilege credentials; do not assume that job activation guarantees an external action occurs exactly once.

Locate the affected process instance and identify the failing element. Inspect error details and relevant variables, then check worker availability, job type, credentials, network reachability and mapping expressions. Do not copy secrets or personal data into support tickets.

Distinguish a business outcome, such as a rejected request, from an operational failure, such as a service timeout. Use explicitly modelled business paths instead of treating every rejection as a technical incident.

Fix the underlying cause before retrying. Confirm whether the external action already happened and reconcile uncertain outcomes. Use your version’s authorised recovery controls and verify that the process reaches its intended state after recovery.

In Tasklist, check assignments, form binding and the permissions of the intended user. Test completion and escalation with representative users. For DMN decisions and FEEL expressions, validate variable types, missing data and boundary values before deployment.

Maintain a short runbook with incident owner, escalation contacts, recovery checks and evidence references. This is an introductory manual, not a production configuration or claim that Camunda has been installed on this website.