Sprite stuck cold: restart 409 and upgrade fails with "could not reserve resource for compute memory"

Questions / Help
1

My Sprite has been unreachable since [time/date].

  • Sprite: purse-view-staging
  • ID: sprite-47e02f13-e416-47d0-94cc-c7b94c4b0dd9
  • Org: ayush-305
  • Version: 0.0.1-rc48 (JuiceFS storage)

What happens:

  • The public URL (https://purse-view-staging-bm4gr.sprites.app) times out with no response.

  • sprite exec -s purse-view-staging -- echo hi hangs.

  • GET /v1/sprites/purse-view-staging returns status: cold, and last_running_at: null.

  • Restarting from the dashboard gives: “The upstream service rejected the restart of purse-view-staging due to a conflict or unmet precondition. Check its status before trying again”.

  • An upgrade with allow_storage_migration: true failed at 2026-10-30T11:50:58Z:

    "type":"failed", "error": "aborted: could not reserve resource for compute memory available to fulfill request on the current host", status: 409

This looks the same as Sprite cannot start; restart returns HTTP 409 , where the host had no capacity and the machine was not moved.

Thanks!

2

:waving_hand: I can confirm that your Sprite’s underlying host was memory constrained at the time. This did prevent it from starting, and explains the failures you saw earlier. The upgrade at 11:50 replaced the original Machine with one on a different host and it is responsive now when I visit your public URL.

I don’t know why your upgrade request returned an error, but what I see on my side shows the upgrade did go through. Your Sprite is now running 0.0.4 and is backed by SBD!

3

I retried the upgrade some time after posting this and the retry worked.

I tried to delete the post but could not figure out a way to delete it.

Thanks for the response :slight_smile: