Delays in starting jobs using the gen3 resource class

Summary

Delays in starting jobs using the gen3 resource class

Between 19:24 UTC on 22 July and 01:50 UTC on 23 July, customers using the gen3 preview machine (Linux VM) resource classes experienced delays and, in some cases, jobs failing to start. We have removed the gen3 resource classes while we address stability issues affecting them. Unfortunately, workflows that we were unable to provision compute capacity for during this period were cancelled. Affected customers should change their resource class from gen3 to gen2 and rerun those workflows. We thank you for your patience while our team worked to resolve this.

• Started

  Jul 22, 2026 · 7:57 PM

• Resolved

  Jul 23, 2026 · 2:05 AM

• Duration

  6h 8m

• Severity

  Minor

• ◐

  Investigating

  Jul 22 · 7:57 PM CircleCI

  We are investigating delays in starting jobs using the gen3 resource class.

• ◐

  Investigating

  Jul 22 · 8:43 PM CircleCI

  We are continuing to investigate elevated wait times affecting customers using gen3 machine executors. Affected jobs may take longer than usual to start. We will provide another update as soon as we have more information to share.

• ◐

  Investigating

  Jul 22 · 11:17 PM CircleCI

  We are continuing to investigate elevated wait times affecting customers using gen3 machine executors. Affected jobs may take longer than usual to start. We will provide another update as soon as we have more information to share.

• ◐

  Investigating

  Jul 23 · 12:31 AM CircleCI

  We have identified the issue and are working with our provider to resolve the availability of gen 3 resources. Gen 1 and Gen 2 are both fully operational.

• ✓

  Resolved

  Jul 23 · 2:05 AM CircleCI

  Between 19:24 UTC on 22 July and 01:50 UTC on 23 July, customers using the gen3 preview machine (Linux VM) resource classes experienced delays and, in some cases, jobs failing to start. We have removed the gen3 resource classes while we address stability issues affecting them. Unfortunately, workflows that we were unable to provision compute capacity for during this period were cancelled. Affected customers should change their resource class from gen3 to gen2 and rerun those workflows. We thank you for your patience while our team worked to resolve this.