Hacker News Front Page · Kyle Kingsbury ·
1 Background
Las pruebas de Jepsen sobre Orkes Conductor detectaron ejecuciones concurrentes de tareas, errores al iniciar flujos de trabajo y tres fallos graves: ejecuciones abandonadas, ejecuciones que desaparecían y flujos marcados como completados sin ejecutar sus tareas. Orkes corrigió los tres fallos graves y varios problemas adicionales en la versión 5.5.6; los retrasos por tareas bloqueadas se consideran coherentes con el comportamiento previsto, mientras que dos errores de la API siguen bajo investigación. El análisis recomienda actualizar y evaluar cuidadosamente la durabilidad de las bases de datos y colas, además de diseñar tareas idempotentes y compatibles con ejecuciones concurrentes.
Orkes Conductor is a workflow orchestration platform which runs user-defined tasks and routes data between them. We tested Orkes Conductor versions 5.3.2 through 5.5.0-rc2 and found unexpectedly frequent concurrent execution of individual tasks, one issue that could significantly delay workflow execution, four minor issues with startWorkflow errors or null IDs, and three serious problems which caused Conductor to permanently abandon or forget workflow executions, or to mark them completed even though their tasks were never run. Orkes has released version 5.5.6 with patches for all serious safety issues, as well as reducing concurrent execution and fixing two startWorkflow errors. The other two startWorkflow errors are under investigation. Delayed execution is consistent with Conductor’s design parameters and need not be addressed. Orkes has published a companion blog post to this work. This research was funded by Orkes Inc., and conducted in accordance with the Jepsen ethics policy.