Public: anyone can read this; the crew can change it.
'Running' means reachable: prove every app and new forge from its public address, alarm when it isn't, and carry websockets into cells #7959
Landed in pull #7981 · 2 runs · run
the crew's run was interrupted: restart, resumes on the next pass — cancelled
Sign in to join this conversation.
new above
offline
The owner created a forge for a new client (Starck) before a first meeting: "the forge says the apps are running but they are not." The routing bug is #7956. The deeper failure is that nothing noticed: Starck's pages said "App · running" for days while every public address answered 404, because "running" only means the process is up inside the cell.
Done when:
/forges, and in MCP./forgeslist each URL with its result. A failure says what failed and raises an alarm instead of showing green.forwardinsrc/spawn/cell.tsis a plain fetch, so an app's websocket upgrade can't reach a cell, whileproxyApphandles it for hosted children. Carry upgrades through, with a test.<app>--<child>--<parent>through the real server and assert the app's own body. The unit test in #7956 fakes the cell; this one doesn't.Also worth deciding:
<app>--<child>.open-platform.sh(a natural mistyping) returns a 200 from the platform's wildcard standby page, so a naive uptime check reads it as healthy. Probes must assert the app's own response, not just the status.