MPG dashboard Users page stuck on "Loading users..." with 5-second polling, `fly mpg users list` returns 504 Gateway Time-out

Questions / Help
1

Hi,

For the past 2–3 weeks the Managed Postgres dashboard has been unable to load the users list, and fly mpg users list from the CLI consistently returns 504 Gateway Time-out after ~60 seconds. Wondering if anyone else is seeing the same thing or has a workaround.

Symptoms

Dashboard — Users tab (every MPG cluster I have)

  • The “Users” page renders the form (“Create a new user”) but the list section below stays on Loading users... indefinitely.
  • The section auto-refreshes about every 5 seconds (network re-fetch, not a full page reload).

Dashboard — Connect tab (only clusters with at least one attached app)

  • The “Connection settings” section under the Connect tab shows the same Loading users... placeholder and re-fetches every ~5 seconds.
  • On clusters with no attached apps, the Connect tab renders fine and does not poll — only the Users tab is broken there.

So the trigger is consistent across the dashboard: anywhere the dashboard tries to enumerate cluster users, it hangs and retries every 5 seconds.

CLI reproduction

fly mpg users list <cluster_id> reproduces the upstream timeout independently of the dashboard, so this looks like an MPG management-plane issue rather than a frontend bug:

4

Tested this with a fresh cluster, fly mpg users list <cluster_id> returned successfully for me, so it’s not a universal MPG issue. Might be specific to your cluster or org. Worth emailing support@fly.io with your cluster ID so they can check the management plane on their end.

5

Hi, one precision: support@fly.io is unmonitored. You can file a ticket directly from Sign in to Your Account · Fly since you have an MPG cluster.

6

Good to know, thanks for the correction!

7

Thanks both — super helpful :folded_hands:

Knowing a fresh cluster in another org works fine is exactly the data point I needed. Definitely org/cluster-specific on the management plane side, not a general MPG thing. And good catch on `support@fly.io` — would’ve sent it there otherwise!

I’ve opened a ticket through the support portal with the cluster IDs and the CLI repro. Will report back here once Fly takes a look, so the resolution is on record for anyone else hitting this.

Cheers!