# Troubleshooting

## Health and diagnostics

```bash
curl -s http://127.0.0.1:8081/healthz          # "ok" when the process and database respond
journalctl -u active-push -f                    # server log
sudo -u active-push active-push device list
active-push probe --url https://mail.example.com/Microsoft-Server-ActiveSync --user jan@example.com
```

`probe` performs OPTIONS, Provision, FolderSync and an Inbox Sync like a phone would. Use `--device-id`,
`--device-type` and `--model` to imitate a specific device. For protocol-level detail add the mailbox to
**ActiveSync -> Logging -> Debug users**; decoded WBXML of every request/response then appears in the log.

## After migrating from Z-Push: "collection ... not found" in the live stream

Phones may keep folder ids created by Z-Push (e.g. `a6271970`). ActivePush answers such a Sync with status 12
(hierarchy changed), remembers the id and sends a FolderHierarchy **Delete** for it on the device's next
FolderSync, so the phone drops the phantom folder and continues with the real ones. If a device still loops
after a few minutes, use **Full resync** in the device dialog or remove and re-add the account on the phone.

## Device keeps asking for the password / HTTP 401

- Wrong password, or the login format does not match IMAP (`general.login_format`, `default_domain`).
- The account or IP may be locked out: **Mailboxes -> Active login lockouts** shows and clears lockouts.
- Check `authentication failed user=... ip=...` lines in the log.

## HTTP 449 or status 142/144

The device has not accepted the current policy. It normally re-provisions automatically. If a device cannot
apply the policy (e.g. it does not support encryption), relax the policy or enable loose provisioning.

## Push is late

Push latency is at most `ping.check_interval_secs` (default 30 s). The reverse proxy must allow long requests
(`proxy_read_timeout 3660s`, buffering off). The **Live** page shows every held Ping and its heartbeat.

## HTTP 503 with Retry-After

The IMAP server was unreachable during the request. Check the **Backends** page test buttons.

## Panel is not reachable

The panel listens on loopback by default; use `ssh -L 9091:127.0.0.1:9091 server`. Starting it on another
address requires `ACTIVE_PUSH_ALLOW_REMOTE_ADMIN=true` (preferably with TLS). Lost password:
`sudo -u active-push active-push admin reset-password`.

## "belongs to uid ..." when running a command

CLI commands must run as the service user so database files keep the right owner:
`sudo -u active-push active-push ...`.

## iPhone still asks for the server address

iOS only runs Autodiscover when it can reach it for the **domain of the e-mail address**. It tries, in order,
`https://<domain>/autodiscover/autodiscover.xml`, `https://autodiscover.<domain>/autodiscover/autodiscover.xml`
and the DNS SRV record `_autodiscover._tcp.<domain>`; a failure is silent and the phone shows the manual
"Server" field. Run the diagnostics on the **Autodiscover** page first (it checks DNS, SRV and every endpoint). Then open **Live** while adding the account:

| What you see | Meaning / fix |
|---|---|
| nothing at all | the request never reaches ActivePush: no DNS record for `autodiscover.<domain>`, a certificate that does not cover that name, or no nginx `server_name` / locations for it. Add the name to the certificate and the three `location` blocks (see `deploy/nginx-active-push.conf`) |
| `401 auth required`, then nothing | iOS did not accept the answer or the TLS certificate; check the certificate chain with a browser |
| `503 backend unavailable` | IMAP login check failed, see Logs |
| `200 ok (host)` but the phone still asks | the returned URL is built from the host the request arrived on (e.g. `autodiscover.example.com`); set **Autodiscover -> Public ActiveSync URL** to the public ActiveSync URL that has a valid certificate |

The e-mail address typed on the phone must be a full address (`user@domain`); the login is derived from it
according to the **Login** settings.