---
order: 1
summary: Why proxility-connector enrol fails, and which cause produces which message.
title: Enrolment rejected
path: troubleshooting/enrolment-rejected
status: published
---

# Enrolment rejected

```
$ proxility-connector enrol --token ptk_...
error: enrolment rejected (token_already_used)
```

The message names the cause. Work from it rather than retrying.

## token_already_used

Enrolment tokens are single-use. Either the connector already enrolled
successfully — check `proxility connector list` before doing anything else — or
someone else used the token.

**If you did not enrol and the token is spent, treat it as a leak.** Revoke the
connector identity and issue a fresh token:

```bash
proxility connector revoke dc1-a
proxility connector create dc1-a
```

## token_expired

Tokens live 60 minutes. Create another. If tokens routinely expire before you
can use them, generate the token as part of the provisioning run rather than
ahead of it.

## token_unknown

A typo, a truncated copy, or a token from a different tenant. Tokens are long;
copy them whole, and avoid shells that wrap.

## identity_in_use

A connector with that name already holds a valid certificate. Names are unique
within a realm. Either pick another name or revoke the existing identity first
if you are genuinely replacing the host.

## network_unreachable

The host cannot reach `api.proxility.scailabs.ai` on TCP/443. This is an egress
problem, not an enrolment problem:

```bash
curl -sv https://api.proxility.scailabs.ai/v1/healthz 2>&1 | tail -5
```

Check the proxy environment too — the connector honours `HTTPS_PROXY` for
control-plane calls but **not** for the data plane, which needs direct UDP or
TCP to the edge.

## clock_skew

Certificate validation needs the host's clock within a couple of minutes of
real time. Fix NTP:

```bash
timedatectl status          # look for 'System clock synchronized: yes'
```

Freshly cloned VM images and long-suspended laptops are the usual culprits.
