Skip to content

History / Command line

Revisions

  • Thirteen events, three scopes, and what each one actually looks like

    @vxnsin vxnsin committed Sep 7, 2026
  • The project file has its own name, and setup asks about the firewall

    @vxnsin vxnsin committed Sep 7, 2026
  • The firewall: four backends, the rollback, and what a rule from the registry may never do

    @vxnsin vxnsin committed Sep 7, 2026
  • Bring the wiki up to 0.2.0: events, webhooks, projects, proxy export, groups

    @vxnsin vxnsin committed Sep 4, 2026
  • Document running programs, holders, doctor, autostart, history, metrics and Docker

    @vxnsin vxnsin committed Sep 2, 2026
  • Document setup, settings, and the Python client Configuration opened with fifteen environment variables and a .env in whatever directory the process started in, which stopped being sensible advice the moment the command could be installed globally. It now opens with `warden setup`, and explains the source order rather than pretending there is only one. The Python client had a table of method names on the API page and nothing that showed anyone how to use it. It gets a page: asking for a port at startup, finding a neighbour, leases and heartbeats, every method, what comes back, which exception means what, and reading the machine with no warden running at all.

    @vxnsin vxnsin committed Sep 2, 2026
  • Bring the wiki up to the current warden Installing by name works now, so the pages say so instead of explaining why it does not. Adds what the fleet gained since they were written: pool and listeners across every node, a name two machines both hold, --node forwarding on register, release and heartbeat, the dashboard's fleet mode, and the new heartbeat command. Records that forwarding takes a person's token rather than the cluster one, and why. Also that the dashboard shows this machine's ports with no warden running at all, which was the last place the two ways of asking the same question still disagreed.

    @vxnsin vxnsin committed Sep 2, 2026
  • Document updates and rolling them across a fleet A page of its own, plus the settings, the endpoints, and a section of troubleshooting built around the messages an update actually prints. States plainly why the hub sends an intent rather than a command, and that restarting the service is the operator's command's job.

    @vxnsin vxnsin committed Aug 31, 2026
  • Document the fleet-wide view warden ls --all, warden get node/service, and the two new endpoints. Also corrects the token model: the cluster token now opens read endpoints too, because a hub asking its nodes what they hold is exactly that. The pages said reading took the human token, which is no longer true.

    @vxnsin vxnsin committed Aug 31, 2026
  • Write the wiki Nine pages: what warden is, how to install it, running it on one machine, seeing and freeing ports without a server, running a cluster, and three reference pages. Plus a troubleshooting page built around the messages warden actually prints, so a reader can search for what they saw.

    @vxnsin vxnsin committed Aug 31, 2026