Machine updates fail with 408 / “concurrent update in progress”

Questions / Help
1

Deploys to one machine keep failing at “Updating machine config”. Same machine every time, intermittent.

  • App: oldworld-wartable-test (org: personal), region arn
  • Machine: 48e32e2c5e0278, one machine, one volume data
  • Method: flyctl deploy, rolling, from GitHub Actions. The image push to registry.fly.io succeeds every time. Nothing else touches the machine.

Errors today (UTC)

  • 12:29:06, 408 (Request ID 01M66BAY68MQW79TZXHRCBB46H-lax)
  • 21:05:57 aborted: machine version changed: concurrent update in progress
  • 21:07:51 aborted: machine is replacing: concurrent update in progress
  • 21:11:14, 21:14:37, 21:20:10: request returned non-2xx status: 408
  • 21:15:12, 21:16:47: machine version changed: concurrent update in progress
  • Final error 21:20:12 (Request ID 01M67A3B0C00DPQN4RY7J4BYMX-iad): failed to update VM 48e32e2c5e0278: aborted: machine version changed: concurrent update in progress

Each update attempt takes 1.5 to 3.5 minutes before it fails. Deploys did succeed at 06:34 and 18:43 today.

Before: on 2026-10-29 a deploy failed with “insufficient memory on the current host”. We moved the machine and volume to another host with volumes fork, which fixed that one.

Tried: re-running the deploy and flyctl’s own retries. Same errors.

Ask: could someone check whether this machine or its host is stuck, and clear any pending “in progress” update? Is there anything we should do on our side?

Thanks!

2

This may no longer be relevant because the machine is now destroyed but for posterity: What you saw was a combination of your machine’s underlying host being at capacity (which made updates slower than usual) and subsequent updates (triggered by flyctl deploy) cancelling each other out. There’s probably some room for improvement in flyctl here. I’ll look into it.