Repository navigation
Closed
Description
Summary
Superset Desktop shows a remote workspace host as offline, but CLI/API checks report the same host as online, healthy, and still
running active workspace sessions.
The Desktop UI displays:
Host is offline
This workspace lives on a device that isn't reachable right now.
Open Superset on that device to bring the workspace back online.
Refreshing the Desktop UI does not resolve it.
Environment
- Superset Desktop:
1.8.8on macOS - Superset CLI on remote host:
0.2.12 - Remote host: Linux cloud VM
- Remote setup: CLI host server started with
superset start --daemon
CLI/API state
Using SUPERSET_API_KEY, both local and remote CLI report the remote host as online:
NAME ONLINE
local-host no
remote-host yes
Remote superset status reports the host service as running:
Team: remote-host running (pid ..., up many hours)
The remote workspaces are also visible via CLI and associated with the remote host.
## Remote host state
SSH to the VM works normally.
The Superset host service and pty daemon are running:
host-service.js
pty-daemon.js
The remote host has active terminal/agent sessions under the Superset pty daemon, so I avoided restarting the host service because that
may interrupt in-flight work.
## Network observations
The VM can resolve and reach:
api.superset.sh
relay.superset.sh
Both respond over HTTPS.
However, when inspecting network connections, I did not see an obvious persistent external connection from the host service process to
the relay. I only saw local loopback connections around the host service port. I’m not sure whether that is expected.
## Local Desktop / auth observation
On the same Mac where Desktop shows the team/workspaces, the local CLI initially said:
Error: Not logged in
Hint: Run: superset auth login (or set SUPERSET_API_KEY)
When I set the API key manually, the local CLI correctly reports the remote host as online.
This suggests there may be a Desktop/local auth/cache/relay state mismatch where API host status and Desktop workspace reachability
diverge.
## What I tried
- Refreshed the Desktop UI
- Confirmed cloud VM is powered on
- Confirmed SSH works
- Confirmed superset hosts list reports remote host online
- Confirmed superset status on remote reports host running
- Confirmed host-service and pty-daemon are running
- Confirmed active remote sessions exist
- Confirmed DNS/HTTPS reachability to api.superset.sh and relay.superset.sh
I did not restart the remote host service because active agents/sessions are running and the docs say stopping the host interrupts in-
flight terminals/work.
## Expected behavior
If CLI/API reports the remote host online and active sessions exist, Desktop should be able to connect to the remote workspace, or show
a more specific diagnostic explaining why the relay/channel is unavailable.
## Actual behavior
Desktop shows the remote host as offline even though CLI/API reports it online and the host service is running.
## Questions
1. How does Desktop determine Host is offline compared to superset hosts list?
2. Is there a non-destructive way to force Desktop to reconnect to the relay?
3. Are there host-service logs or relay diagnostics I should collect?
4. Is restarting the host service currently the only supported recovery path?
5. If restart is required, is there a way to avoid killing active terminal/agent sessions?
Metadata
Metadata
Assignees
Labels
No labels
Activity
saddlepaddle commentedon May 11, 2026
Apologies for the disruption! We'll have the relay back up in a moment, I was testing expanding the regions globally.
I'll post here when it's resolved
saddlepaddle commentedon May 11, 2026
Also out of curiosity are you in Europe? how is your latency currently?
fix: solve #4407 — distinguish unreachable host from offline in Desktop
saddlepaddle commentedon May 11, 2026
This should be back online now!