Jepsen: Orkes Conductor 5.3.2
Web page · https://jepsen.io/analyses/orkes-conductor-5.3.2
First seen by Oriel on .
AI summary
Jepsen tested Orkes Conductor 5.3.2 through 5.5.0-rc2 and found concurrent task execution, workflow stalls, API errors, and serious execution-loss risks addressed partly in 5.5.6.
- Healthy clusters occasionally ran the same task concurrently; with short tasks and roughly 1,100 workflows per second, this occurred about once every six hours.
- A sweeper time-of-check/time-of-use bug re-enqueued tasks after workers began them, exacerbating duplicate delivery; Orkes fixed it in version 5.5.6.
- With Postgres and Redis, faults could leave workflows appearing RUNNING for the default one-hour response timeout before tasks were retried.
- startWorkflow occasionally returned null IDs, syntax errors during partition rotation, duplicate-key errors in 5.5.0-rc1, or connection failures after server termination.
- Testing used Jepsen, the Conductor Java SDK, ten-machine clusters, single-task workflows, process failures, and network partitions across servers, databases, queues, and workers.
#workflow orchestration #distributed systems #postgres #redis
Generated by gpt-5.6-sol. Check the original for details.
Metric trends
Last 30 days · Up to 120 recorded samples per source and metric, with reposts kept separate · First and latest retained
Points
Score reported by the source.
- Hacker News: 59, latest observation ; +44 over observed period since
- Lobsters: 33, latest observation ; +30 over observed period since
Comments
Comment count reported by the source.
- Hacker News: 3, latest observation ; +3 over observed period since
- Lobsters: 1, latest observation ; +1 over observed period since
Aliases
| Source | External ID | URL |
|---|---|---|
| Hacker News | 50165904 | https://jepsen.io/analyses/orkes-conductor-5.3.2 Read discussion |
| Lobsters | 42zgmq | https://jepsen.io/analyses/orkes-conductor-5.3.2 Read discussion |