Docs · Pushbox, Tokenmail and Ping

Ping: uptime monitors and status pages

froog checks your websites, ports, hosts and cron jobs, emails you when one goes down and when it's back, and keeps a public status page for them that you can link to. 5 monitors checked every 5 minutes are free; every other monitor is $1 a month (those beyond the 5, and any checked faster, down to every 30 seconds). Status pages, custom domains, incidents and subscribers are included. See froog Ping.

$ froog ping add https://example.com                     # http, every 5 minutes (free)
$ froog ping add https://api.example.com/health --every 30s --keyword '"ok":true'
$ froog ping add db.example.com:5432 --name db           # tcp: the port accepts a connection
$ froog ping add vpn.example.com --name vpn               # icmp: the host answers ping
$ froog ping add --name nightly-backup --every 1d --grace 30m   # a heartbeat: prints the URL your job calls
$ froog ping ls · show NAME · check NAME · edit NAME --every 1m · pause NAME · resume NAME · rm NAME
$ froog ping page create acme --title Acme --monitor example.com=Website --monitor db=Database
$ froog ping page domain acme status.acme.com            # prints the CNAME and TXT records to add, then checks them
$ froog ping incident open acme "Slow logins" --impact minor --monitor example.com
$ froog ping incident update acme inc_… identified --body "A bad deploy; rolling back"
$ froog ping incident open acme "Database upgrade" --from +1440 --until +1500   # maintenance, minutes from now

What's checked. http: the URL answers with a status in expect_status (default 200-399; redirects are followed, 5 at most, unless expect_status names the redirect), with keyword in the first 1 MB of the body (or, with keyword_absent, not in it), within timeout_secs (default 10). For https, a bad certificate fails the check, and you're warned cert_warn_days (default 14) before it expires. tcp: host:port accepts a connection. icmp: the host answers one ping. heartbeat: your job calls its URL (GET or POST, no API key: the URL is the secret) at least every interval_secs plus grace_secs; <url>/fail?msg=… reports a failure at once. Checks come from froog's servers, so targets must be public: a private address (10.0.0.1, 192.168.…, localhost, one of froog's), or a name that resolves to one, is refused with a 400 when you add or edit the monitor, and still fails a check if its name later resolves to one. If froog itself can't send a check (ICMP not allowed on its server), the check doesn't run and doesn't count against the monitor: last_error says so and froog ping check answers 503 check_unavailable. Set the certificate warning with --cert-warn DAYS (0 turns it off), or on the monitor's page in the dashboard.

Down and up. A monitor is down after confirmations failed checks in a row (default 2; a failed check is retried within 20 seconds to confirm it), and a heartbeat as soon as its deadline passes. Going down opens an outage, emails you and sends ping.monitor.down to the webhooks that list it; the first good check ends it with ping.monitor.up and another email. A certificate about to expire sends ping.cert.expiring. Each event carries the monitor and, for down and up, the outage. Turn a monitor's emails off with --email off, or all of them in Alerts (ping). Checks are kept 7 days (GET /v1/ping/monitors/{monitor}/checks), daily totals and outages longer (uptime over 24 hours, 7, 30 and 90 days).

Status pages. A page lives at https://status.froogusercontent.com/<slug> and shows the monitors you pick (with the names you give them) with 90 days of daily uptime bars, current incidents, scheduled maintenance and the last 30 days of incidents. It has no JavaScript at all. <page>/json is the same as JSON for your own scripts (CORS open), and <page>/badge.svg a status badge for a README. With auto_incidents (on by default) a monitor going down opens an incident on every page that shows it, resolved when it's back up; so does a monitor that's already down when you add it to a page (one per outage: resolve it by hand and it stays resolved), and taking a monitor off a page resolves its incident there. Maintenance that covers a monitor holds its incident back until the maintenance ends. Changing a page's slug keeps the old address working: it redirects (301) to the new one. A page keeps its last 5 slugs, and no other account can take them while the page exists; you can take one back yourself.

Your own domain. Set custom_domain (froog ping page domain acme status.acme.com) and add two DNS records: a CNAME from your name to status.froogusercontent.com (serves the page) and a TXT record at _froog-status.<your name> (proves it's yours; the value is shown). Once the TXT record is found the page is served there over HTTPS, with a certificate made on the first visit, and the status.froogusercontent.com address redirects to it. froog looks again hourly; a domain whose TXT record stays gone for 7 days stops being served.

Incidents and maintenance. An incident has a title, an impact (minor, major, critical), the monitors it affects and updates, each with a status (investigating, identified, monitoring, resolved). Maintenance has a window (starts_at, ends_at) and the statuses scheduled, in_progress, completed; during its window the monitors it names show as under maintenance.

Subscribers. Visitors can subscribe to a page by email (subscribers_enabled, on by default). They confirm from a link first, and every incident update that notifies (the default; --quiet or "notify": false doesn't) is emailed to them with a one-click unsubscribe link. An incident or maintenance opened quietly stays quiet: its later updates (going in_progress, completed) aren't emailed either, unless one says "notify": true (froog ping incident update … --notify). froog ping subscribers acme lists them.

Paying. Free monitors are your first 5 that check every 5 minutes or slower. Every other one is a seat on a Stripe subscription of its own: the first paid monitor opens Checkout (froog ping add does it for you, or POST /v1/ping/billing/checkout; without a terminal, as from an agent, or with --no-wait, the CLI prints the link for a person to open and exits with an error instead of waiting), later ones are added to it at once, charged for the rest of the month. Deleting a paid monitor keeps its seat until the month ends; with none left the subscription ends then. While the subscription isn't paid, paid monitors aren't checked (free ones still are). froog ping billing shows where you are.