Maintain
Health & alerts
Integration Health answers one question continuously: is Wrenda still correctly in front of this domain? Seven checks roll up into a status, the status is re-checked every hour, and you are told when it gets worse.
The seven checks
Domains → your domain → Integration Health lists each check with a detail line, the overall pill, when it was last checked, a Run live check button and — when there is something Wrenda can repair — a Fix button.
| Check | What it verifies | If it fails |
|---|---|---|
| DNS CNAME | The domain resolves to proxy.wrenda.ai (looked up over DNS-over-HTTPS). | Traffic is not reaching Wrenda. The record was removed, changed, or is proxied (orange cloud) on Cloudflare. Re-add it as DNS only. |
| DNS DCV | The certificate-validation CNAME is present. | The current certificate keeps working, but renewal (every 90 days) will fail. Re-add the _acme-challenge CNAME from the wizard. |
| DNS TXT | The ownership TXT record is present. | Wrenda can’t re-prove ownership if the hostname needs to be re-created. Re-add the TXT record. |
| Certificate | The Custom Hostname certificate is active; shows the expiry date. | Pending: validation hasn’t completed — usually the DCV record. Expired or errored: see Troubleshooting → certificate stuck pending. |
| Worker route | The edge route that sends the hostname’s traffic into Wrenda exists. Skipped for CDN and web-server integrations. | Requests reach Wrenda’s edge but bypass the proxy. Fix re-creates the route. |
| Origin | The origin answers, with its HTTP status and latency. | Nothing can be served. Origin down, wrong origin URL, Host header needed, or an origin loop. Check the status code in the detail line. |
| Traffic | Recent crawler activity: last crawler seen (and its user agent), last optimization, and the 24-hour cache-hit ratio. | A warning, not an outage: no crawler has visited recently, or nothing is being optimized. Check the mode (Shadow/Paused) and the rules. |
How the overall status is derived
- Down — any failure in DNS CNAME, certificate, worker route or origin. Crawlers are not getting served correctly.
- Degraded — any other warning or failure (DCV, TXT, traffic). Serving works today but something needs attention.
- Healthy — every check is ok or skipped.
- Unknown — not checked yet.
The dashboard home shows the same pill per domain next to its setup progress.
Live checks and the hourly sweep
Run live check runs all seven checks now and stores the report. Independently, an hourly sweep re-checks up to 50 verified domains, oldest-checked first, so every domain is refreshed regularly without you doing anything. The stored report is what the dashboard, the home page pill and the alerts use.
Regression alerts
When a sweep sees the overall status go from healthy to degraded or down, Wrenda creates an in-app notification, records it in the alert history under Settings → Alerts (type integration health), and emails the tenant's admins. When the status recovers you get a recovery email. Both are de-duplicated per day, so a flapping origin produces one regression and one recovery per day, not a flood.
These sit alongside the existing alert rules (usage limits, activity drops, error rates, cache performance) and Citations drift alerts.
Lifecycle emails
Short, one-link emails to tenant admins at the moments that matter:
| Sent when | |
|---|---|
| Verified and serving | The domain passes verification for the first time (or the health check first sees the CNAME and route both ok). |
| Certificate active | The Custom Hostname certificate becomes active. |
| First AI crawl | The first request from an AI crawler is logged for the domain. |
| Health regressed | A sweep sees healthy → degraded/down (once per day). |
| Health recovered | A sweep sees the status return to healthy (once per day). |
| Allowance reached | The plan’s monthly uncached allowance is used up. Crawlers get cached copies or your original pages from this point; nothing new is optimized. |
| Daily reminder | Every day while usage stays above the plan, counting down the 30-day warning window. |
| Service reduced | The 30-day window has ended: every request now passes straight through until you upgrade. |
| Full service restored | An upgrade covers your usage; optimization and pre-rendering resume and the reminders stop. |
The allowance emails are about billing, not integration health: your site is never affected by them. Details of what happens at the limit are in Cache & purge.
The Fix button
Fix appears when the failing condition is one Wrenda can repair itself. It will:
- Re-create the worker route if it is missing.
- Re-run verification if DNS is correct but the domain is still marked unverified.
- Re-run the health check and show the new report, together with the list of actions it took.
Fix cannot change your DNS, your origin or your certificate validation records — those come back to you as instructions in the check's detail line.
CDN and web-server integrations
For these, Wrenda isn't in DNS, so the DNS and certificate checks reflect only what Wrenda can see and the worker-route check is skipped. The origin and traffic checks are the useful ones: a traffic warning means your server has stopped forwarding bot requests to /proxy/. See verifying a snippet.