Standardize fleet site agents on Pangolin CLI in Docker #44
Loading…
Reference in a new issue
No description provided.
Delete branch "pangolin-cli-site-agents"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Replaces the fleet's mixed agents with the pattern Pangolin now recommends: all three phd-* hosts run
fosrl/pangolin-cliin Docker through the Cumulus stack, so the agent auto-starts and restarts with everything else instead of being a manually-launched host process.Two roles per machine, both the same image:
pangolin-site(site mode,SITE_ID/SITE_SECRET) — accepts inbound overlay connections and forwards resources tolocalhostports. Beelink and razer swapfosrl/newtfor this; credentials carry over as-is since the CLI accepts existing site ID/secret pairs (verified live on razer). Tower already ran this container.pangolin-client(machine-client mode,CLIENT_ID/CLIENT_SECRET) — the only component that creates a host's outbound routes and.internalDNS resolution; this replaces the manualpangolin up clientprocesses that had to be started by hand after every boot.Also updates Makefile help text and the
.env.examplecomments to name both credential pairs.Rollout note: beelink's container swap happens after merge, via
git pull+make up phd-beelinkon that host; each host also needs its machine-client credentials added to its.env(*_CLIENT_ID/*_CLIENT_SECRET), freshly generated from the dashboard.