One view across your fleet
Connect to multiple HAProxy stats endpoints from one config file. Compare health and traffic without hopping between admin pages.
A focused, read-only dashboard for HAProxy frontends, backends and servers. One native binary, a small config file, and the visibility your fleet deserves.
| Server | Status | Sessions | Traffic in / out | Last activity | Last check |
|---|---|---|---|---|---|
| web-01 | UP | 42 / 200 | 1.8 GB / 6.2 GB | 12s ago | Layer7 OK 200 |
| web-02 | UP | 38 / 200 | 1.6 GB / 5.8 GB | 4s ago | Layer7 OK 200 |
| web-03 | DOWN | 0 / 200 | 0 B / 0 B | No session yet | Connection refused L4 |
| web-04 | UP | 31 / 200 | 1.4 GB / 4.9 GB | 1m ago | Layer7 OK 200 |
Proxyglass turns HAProxy’s stats feed into an operational overview, and deliberately stops at visibility.
Connect to multiple HAProxy stats endpoints from one config file. Compare health and traffic without hopping between admin pages.
See frontends, backends and individual servers together, with sessions, byte counters, status and health-check details.
Check each server’s last session and health check. Spot stale or unhealthy pool members before they surprise you.
Filter by proxy, frontend, backend or server. Sort the data and keep your preferred refresh interval in a browser cookie.
Proxyglass only reads stats. There are no buttons or API calls to disable, drain or otherwise change HAProxy state.
The UI assets ship with the binary. No CDN, hosted fonts or third-party runtime calls are needed to open the dashboard.
Use locally generated bearer tokens or connect an OIDC provider. An optional group allow-list keeps access aligned with your identity provider.
Run a static Linux binary under systemd with config and environment in familiar /etc locations. No container platform is required.
Use the optional image and Nomad example if that is how your infrastructure is operated. The same config and read-only behavior apply.
Proxyglass sits alongside HAProxy, polls its read-only CSV stats endpoint, and serves the dashboard to your team.
Use a dedicated read-only account and the ;csv stats endpoint.
Put endpoint URLs and credentials in Proxyglass’ config file.
Start the binary, sign in, and filter the parts of the fleet you need.
# Proxyglass listens behind your existing nginx
server:
listen: "127.0.0.1:8080"
default_refresh: 5 # 0 disables refresh
targets:
- name: edge-01
url: https://edge-01/stats;csv
username: stats-reader
password: <secret>
- name: edge-02
url: https://edge-02/stats;csvProxyglass is a small Go service, not a platform you have to operate.
Install a static Linux binary, create a service account, and let systemd keep the dashboard running. Nginx remains responsible for TLS.
/etc/proxyglass/Prefer containers? Use the same config in a Docker image and schedule it as a small Nomad service.
Air-gap friendly by default. Dashboard JavaScript, styles and logos are bundled locally. For OIDC or external HAProxy targets, standard proxy environment variables are supported.
Start with one HAProxy stats URL. Add the rest when you’re ready.
Get Proxyglass Or start with the config example ↑Proxyglass is read-only by design.