# InfraNest monitoring probes InfraNest is an infrastructure management platform: domains, DNS, servers, certificates and monitoring in one place. Its monitoring probes check websites and services for InfraNest customers. If you found **`InfraNest-Monitor`** in your logs or your firewall, a request came from one of those probes: an InfraNest customer asked us to check that address, and our probes check it on a schedule from the regions the customer picked. This page says what the probes do and do not do, how to recognise them, where they connect from, and how to reach us. ## What a probe does - **It checks one address, on a schedule.** A customer adds a monitor for an address they run or are responsible for. Each region the customer picked checks it at the interval they chose: every 30 seconds at the most often, once a day at the least. A check is one request: an HTTP(S) request, a TLS handshake, a TCP connection, a ping or a DNS lookup, depending on what is being monitored. - **A failure may be checked again.** When a check fails, another of our probes may repeat it once, straight away, to tell a problem with your site apart from a problem with our probe's route. A probe may also make one check now and then to measure which region is closest to the address. Nothing else is added. - **It keeps only the result.** Whether the address answered, how long each step took, the status code, the certificate, and for some checks a short part of the answer as evidence (for example, whether a word the customer asked about was on the page). It does not store your pages. A result and its excerpt are kept for three days; after that only hourly figures remain (response times, counts, uptime), and after 30 days daily ones. ## How often your site is checked - **Each monitor at its own interval, per region.** A customer's monitor checks your address once per interval — every 30 seconds at the most often — from each region they picked, plus at most one repeat after a failure. - **Never a burst from one probe.** A probe runs at most 4 checks against one host at a time, and starts two checks of the same host at least 300 milliseconds apart. - **Several customers may monitor the same site.** Each has their own monitor, so these limits are per probe and per monitor, not a total for your site. The `X-InfraNest-Check` value tells them apart. ## What a probe does not do - **It does not crawl.** It requests only the address the customer entered — and, for an HTTP check, the redirects that address answers with. It does not follow links, load images, scripts or styles, run JavaScript or submit forms. That is why `robots.txt` does not apply: it is written for crawlers that discover pages, and a probe discovers nothing. - **It does not choose its own targets.** We never monitor an address on our own initiative. Every check belongs to a customer's monitor. - **It does not try to get past your protection.** It does not pretend to be a browser, solve challenges or rotate addresses. When your bot protection or firewall turns a check away, the check records that it was turned away, and the customer is told how to allow our probes. ## How to recognise a check Every check from our probes carries these, so you can find our requests in your logs and match them to your rules: | | Value | |---|---| | **User-Agent** | `InfraNest-Monitor/ (+https://developers.infranest.io/probes)`, with the release in place of ``. **Every request carries it.** A customer may set a User-Agent of their own for their monitor; it is sent with ours appended — `TheirAgent/1.0 InfraNest-Monitor/ (+https://developers.infranest.io/probes)` — so match ours anywhere in the value, not only at the start. | | **Check header** | `X-InfraNest-Check: .`, on every request. A customer cannot change or remove it. Quote it when you write to us, and we can find the monitor. | | **Address** | One of the addresses below. | | **Reverse DNS** | The names listed beside the addresses, where an address has one that resolves back to it. | | **Signature** | `Signature-Agent`, `Signature-Input` and `Signature` headers — see [Signed requests](#signed-requests-web-bot-auth). | A User-Agent and a header can be copied by anyone. To be sure a request is ours, check one of these: 1. **Its signature** — see [Signed requests](#signed-requests-web-bot-auth). This works from any address. 2. **Its reverse DNS.** Look up the name of the address the request came from (`dig +short -x 203.0.113.7`). It must end in `.probe.infranest.io` or be `prod-app.infranest.io`. Then look that name up (`dig +short gra.probe.infranest.io`) and check that it gives the same address back. A name that does not resolve back proves nothing: anyone can set a reverse name. 3. **Its address**, against the list below. The list below holds the probes that check every customer's monitors. A few probes check only one customer's monitors and are not on it. A request from one of those is still ours when it is signed or its reverse DNS checks out as above. ## Signed requests (Web Bot Auth) Every HTTP request a check makes — each redirect it follows included, a WebSocket monitor's handshake and a gRPC monitor's health call too — is signed with [Web Bot Auth](https://datatracker.ietf.org/doc/draft-meunier-web-bot-auth-architecture/): HTTP Message Signatures ([RFC 9421](https://www.rfc-editor.org/rfc/rfc9421)) made with an Ed25519 key that belongs to the probe. A request carries: ``` Signature-Agent: "https://api.infranest.io" Signature-Input: sig1=("@authority" "signature-agent");created=1762632000;keyid="…";alg="ed25519";expires=1762632060;nonce="…";tag="web-bot-auth" Signature: sig1=:…: ``` - **The signature covers** the host the request was sent to (`@authority`) and the `Signature-Agent` header, and is good for 60 seconds from `created`. - **`keyid`** is the key's JWK SHA-256 thumbprint ([RFC 7638](https://www.rfc-editor.org/rfc/rfc7638)). - **Our keys** are published at [`https://api.infranest.io/.well-known/http-message-signatures-directory`](https://api.infranest.io/.well-known/http-message-signatures-directory) — a JSON Web Key Set served as `application/http-message-signatures-directory+json`, with one signature per key over the response (`tag="http-message-signatures-directory"`). It lists the current key of every one of our probes — those on the list above and those that check only one customer's monitors, new probes included; a key we have withdrawn is removed at once. CDNs and firewalls that verify signed bots read the directory themselves. Cloudflare lists our checks as the verified bot [`InfraNest-Monitor`](https://radar.cloudflare.com/bots/directory/infranest-monitor) (*Monitoring & Operations*), recognised by this signature. If yours does not, you can verify a request yourself: rebuild the signature base from `@authority` and `Signature-Agent` as RFC 9421 §2.5 lays out, find the key by `keyid` in the directory, and check the Ed25519 signature. A request without a signature is still one of ours when it comes from our addresses: a probe whose clock has drifted, or that is not able to sign, sends its checks unsigned rather than not at all. Signing is in addition to our addresses, never instead of them. ## Where the probes connect from The same list, for scripts and firewalls: - **JSON**: [`https://api.infranest.io/api/probes.json`](https://api.infranest.io/api/probes.json) — every probe with its location, region, provider, addresses and their reverse DNS names, whether it is new, and the User-Agent and check header above. - **Plain text**: [`https://api.infranest.io/api/probes.txt`](https://api.infranest.io/api/probes.txt) — one address per line, IPv4 first, then IPv6. Both lists are public and need no token. They are cached for five minutes and carry an `ETag`, so reading them often costs nothing: re-read them every hour or so, and replace your copy when they change. The addresses change rarely. `updated_at` in the JSON says when the list last changed. ### Fields of the JSON list | Field | Meaning | |---|---| | `updated_at` | When the list last changed: a probe added, removed or moved. | | `version` | A fingerprint of the addresses. It changes exactly when an address is added or removed. | | `user_agent`, `check_header` | How our checks identify themselves (above). | | `addresses` | Every address, IPv4 first, each once — the same as the plain-text list. | | `probes[].name`, `location`, `region`, `country`, `provider` | Where the probe is, and whose network it runs on. | | `probes[].ipv4`, `ipv6` | The addresses its checks connect from. Either can be `null`. | | `probes[].ipv4_hostname`, `ipv6_hostname` | The reverse DNS name of the address, when it resolves back to that address; otherwise `null`. | | `probes[].in_probation` | `true` for a new probe we are still testing. Its checks already reach your site, but their results count for nothing yet. It is listed so an allowlist built from this list is ready before it counts. | Fields are only ever added, never renamed or removed, so a script written against this list keeps working. ## Allowing our probes If your firewall, WAF or bot protection blocks our checks, the customer's monitor cannot see your site, and they are told so. Allow our probes **by address**, from the lists above — a rule on the User-Agent alone lets in anyone who copies it. [Allow InfraNest's probes](https://infranest.io/docs/allow-our-probes) has the steps for Cloudflare, AWS WAF, Akamai, Fastly, Vercel, Imperva, Sucuri, cPanel and Imunify360, fail2ban and nginx. ## Contact - **A probe is checking a site of yours that you did not ask for, or it is causing you a problem:** write to [abuse@infranest.io](mailto:abuse@infranest.io) with the time of a request and its `X-InfraNest-Check` value. That value names the monitor, so we can find it and reply. - We look into every report. If our checks are causing your site a problem, we pause them straight away. - If you run the site and do not want it monitored, we stop the checks once you have shown that it is yours. We tell you how; it is usually a DNS TXT record. - Blocking our addresses, or answering our checks with a 403, stops them reaching you at once. A probe never tries to get around a block. - **A security issue:** [security@infranest.io](mailto:security@infranest.io) — see [Report a security vulnerability](https://infranest.io/docs/report-a-vulnerability). - **Anything else:** [contact us](https://infranest.io/contact). InfraNest customers: [support](https://infranest.io/support).