All systems operational
The supervisor checks every running job about every 30 seconds. Last check 40 s ago. A full check of all jobs takes 0.8 s.
Supervisor uptime, last 30 days · 99.02%
16
Jobs interrupted by a machine failure · last 30 days
11
Jobs resumed after a failure · last 30 days
1 min
Median time to resume · last 30 days
2.5 min
90% resumed within · last 30 days
12
Jobs completed · last 30 days
Only aggregate numbers: no client, job or machine details. Time to resume runs from the moment a failure is noticed until the job reports from the new machine, including image and model download. Our own test runs and jobs stopped by an error in their code or image are not counted.