---
id: fleet/an-orchestrator-that-acts-on-every-server-at
title: "An orchestrator that acts on every server at once"
theme: fleet
status: open
created: 2026-08-20
date_source: assumed
order: 1
---

**An orchestrator that acts on every server at once** — pull and update every platform, restart a
process everywhere, report which host is behind. The list is the input; this is the thing that
reads it.

Three facts from writing the list decide the design, and all three are verified: the checkout
**path differs per host** (take it from PM2's `pm_cwd` for the `officer` process, never a
hard-coded path); `alpha` carries a **second, stale checkout** so pulling the wrong one succeeds
while changing nothing; and the **git remotes differ** — one host pulls over HTTPS and one points
at a different Gitea altogether. An orchestrator built on "ssh, cd to the same path, git pull"
is wrong on three counts before it starts.

Open: whether this is a script over SSH, an Officer screen, or a chat capability; and whether
`headscale` and `omega` — checkout present, nothing running — are targets at all. Note the
platform already has an **Update button** for a single server (`03abdeb2`); this is that, fanned
out, and it should probably share the code rather than reimplement the pull.