Sprite cannot start; health check unhealthy and restart returns 502

Questions / Help
1

One of our three Sprites has been unreachable since approximately 2026-09-24 05:34 UTC. The other two respond normally.

  • Metadata GET succeeds and reports Cold.
  • The SDK health endpoint GET /check reports status unhealthy, reason failed to start machine (latest 06:01:55 UTC).
  • Both service-status requests time out after 30–35 seconds.
  • A harmless /bin/true exec returned HTTP 502.
  • The dashboard Restart reports upstream restart failed; one POST /restart returned the same error, HTTP 502, at 06:01:52 UTC.
  • Checkpoint listing also times out, so we cannot verify available recovery points.
  • These failures reproduce from outside our Cloudflare application.

We found a similar resolved report: Sprite unreachable; dashboard restart returns “upstream restart failed” . Could the Fly team inspect the backing machine and help restore access while preserving the current filesystem and checkpoints? Please contact us before any rollback or action that could discard writes.

Following the identifier-hash approach used in that thread:

  • Organization slug MD5: cf8e72832ea0801999c2e2f3689f74da
  • Sprite name MD5: 35f3d89ba758eac44439eab3af1de20f

We have not deleted, restored, or replaced the Sprite.

2

Hi there, I checked and the Sprite you mentioned is alive and kicking, it’s responsive to requests and can exec commands. Can you verify if it revived for you as well?