3.4 KiB

Raw Permalink Blame History

Clavitor POP Deployment — Handoff for Hans

From: Johan / James Date: March 7, 2026 Status: Binaries built, download endpoints added, ready for your rollout

What's ready on HQ (noc.clavitor.com)

Two new binaries in /home/johan/clavitor-dashboard/:

File	Arch	Size
`clavitor`	linux/amd64	~18MB
`clavitor-arm64`	linux/arm64	~18MB

Download endpoints added to the dashboard (need rebuild):

http://185.218.204.47:8080/download/clavitor
http://185.218.204.47:8080/download/clavitor-arm64

To activate: rebuild dashboard-go and restart the service.

What's new in the binary

Built-in telemetry. When launched with these flags, the vault POSTs system + vault metrics to HQ every N seconds:

--telemetry-freq=60
--telemetry-host=http://185.218.204.47:8080/telemetry
--telemetry-token=<your-choice>

Also works via env vars: TELEMETRY_FREQ, TELEMETRY_HOST, TELEMETRY_TOKEN. Without flags, telemetry is off — no behavior change for self-hosters.

Payload (JSON POST):

{
  "version": "0.1.0",
  "hostname": "virginia",
  "uptime_seconds": 3600,
  "timestamp": "2026-03-06T10:00:00Z",
  "system": {
    "os": "linux", "arch": "arm64", "cpus": 2,
    "cpu_percent": 12.5,
    "memory_total_mb": 1024, "memory_used_mb": 340,
    "disk_total_mb": 8000, "disk_used_mb": 1200,
    "load_1m": 0.3
  },
  "vaults": {
    "count": 0, "total_size_mb": 0, "total_entries": 0
  },
  "mode": "hosted"
}

What needs doing

1. Telemetry inbox on the dashboard

The dashboard doesn't have a /telemetry handler yet. You'll want to add one that:

Accepts the JSON payload above
Stores it (SQLite, or just update the existing nodes table)
Feeds into the status page

This is your call on how to wire it in — you know the dashboard code best.

2. Wipe the status DB

Johan wants the status.db wiped clean and rebuilt with only the three live nodes:

Node ID	Name	Region	IP
`hq-zurich`	HQ — Zürich	Hostkey / CH	185.218.204.47
`virginia`	Virginia	us-east-1	?
`singapore`	Singapore	ap-southeast-1	47.129.4.217

Important: The current "virginia" POP is tagged us-east-2 with IP 3.145.131.247 — that's Ohio (Dublin). Johan does NOT want Ohio. Please confirm:

Was this already moved to us-east-1 (actual Virginia)?
If not, we need to spin down Ohio and deploy in us-east-1.

The planned nodes (london, frankfurt, tokyo, etc.) can stay in the seed data as "planned" but shouldn't be in the live status rotation until deployed.

3. Deploy clavitor to the two AWS POPs

Each POP needs:

The clavitor binary (arm64 for t4g.micro)
A systemd service with telemetry flags pointing to HQ
Port 1984 open
DATA_DIR for vault storage

You already have deploy-pop.sh and SSM access — adapt as you see fit. The clavitor binary replaces nothing; it runs alongside the existing v1984-agent (or you can consolidate, since clavitor now reports its own metrics).

4. Infrastructure overview (for clarity)

Server	Role	Location
zurich.inou.com	Kuma, security checks, shared git	Hostkey Zürich
noc.clavitor.com	Dashboard, status page, marketing site, HQ	Hostkey Zürich
virginia POP	Clavitor hosted node	AWS us-east-1 (confirm!)
singapore POP	Clavitor hosted node	AWS ap-southeast-1

Questions? Ping Johan or ask James in the next session.

3.4 KiB Raw Permalink Blame History