Skip to main content

Missing a check? Tell me, I'll add it.

· 5 min read
Florent Clairambault
Maintainer of SolidPing

Pick any monitoring tool, run it for a week, and you will eventually find the one thing it can't check. Maybe it's a TCP port speaking a non-standard protocol. Maybe it's a JSON field three levels deep in a GraphQL response. Maybe it's whether your Minecraft server still has any players online.

For SolidPing, my answer to all of the above is the same: tell me, and I'll add it.

This is the policy, written down once so I can stop hedging in DMs:

If a non-trivial number of people on Earth want to monitor a thing, SolidPing should be able to monitor it.

That's the rule. The rest of this post is the asterisks attached to it.

What's already in the box​

There are currently 43 check types shipped, which is more than I expected to be writing on a post. The full list lives in the docs, but here's the speed-tour:

  • Network — HTTP/HTTPS (with body matching, headers, JSON paths, custom methods), TCP, UDP, ICMP ping, DNS records, WebSocket.
  • Security — TLS certificate expiry and chain validation, domain expiration via WHOIS (because forgetting to renew a domain is the most expensive kind of outage).
  • Databases — PostgreSQL, MySQL/MariaDB, Microsoft SQL Server, Oracle, MongoDB, Redis. Each with an optional sentinel query so you can probe more than just the TCP port.
  • Email — SMTP (with STARTTLS), IMAP, POP3.
  • Remote access — SSH, FTP, SFTP.
  • Messaging & streaming — gRPC (using the standard health protocol), Kafka, RabbitMQ (AMQP), MQTT.
  • Infrastructure — SNMP polling, Docker daemon health, game-server queries via the Steam protocol.
  • The fun ones — Heartbeats (for cron jobs, and the inverse case where the service pings us), arbitrary JavaScript scripts, and a full headless browser for SPAs, login flows, and visual checks.

If you came from another tool, most of this isn't going to surprise you. The boring 90% has to be solid before anything fun matters.

The stuff that's fair game​

Here is where it gets interesting. None of these are in SolidPing yet, and all of them are exactly the kind of request I'd be happy to take. (Three items from my first list have shipped since I wrote it: Minecraft server queries, NTP and ClickHouse. Here's what's left.)

  • Cryptocurrency node sync — block height behind by N blocks? Bitcoin, Ethereum, anything with a JSON-RPC.
  • GitHub / GitLab repo activity — "tell me if my own repo has gone 30 days without a commit, because honestly that's a problem too."
  • RSS / Atom feed freshness — for the people who still publish blogs (hi).
  • Email reputation — DKIM, SPF, DMARC, DNSSEC validation. The regular DNS check covers records; this would cover whether they actually mean what you think they mean.
  • Lighthouse performance scores — alert me if the homepage drops below 90.
  • BGP peer status, LDAP bind, RADIUS auth — for the network folks who still need their tools to be useful in 2026.
  • Elasticsearch cluster health, Cassandra — the database list keeps growing because the database world keeps growing.
  • CalDAV, WebDAV, IPFS gateways, Tor hidden services — protocols that nobody asks for until one person desperately needs them.
  • Modbus, OPC UA — for the industrial-control crowd who deserve good tools too.

I could keep going. I will keep going, in fact, every time someone opens an issue.

The fine print​

There's exactly one filter, and it's gentle: "a little bit used" means I won't add a check for a protocol that hasn't shipped a release in fifteen years and has nine known users (sorry, [insert nostalgic protocol of your choice]). But the bar is low. Genuinely low. If you, two friends, and a Stack Overflow thread care about it, that's enough.

The reason for the filter isn't gatekeeping — it's that every check is a small surface I have to maintain, document, and not break. Adding fifty-seven checks I don't understand is how a monitoring tool turns into a museum.

How to ask​

The fastest path is a GitHub issue — or a discussion if you're not sure whether your idea is unhinged or brilliant. Whichever it is, let me decide. In the request, please describe:

  • The protocol or service.
  • What "up" means for it. (This is the actual hard part.)
  • A link to a spec, an existing client library, or any working example.

If a Go library already exists for the protocol, the ETA is usually a weekend or two. If I have to write the parser from scratch, give me a few weekends. Either way, you'll get an answer — even if the answer is "this doesn't quite fit, here's why."

Pull requests are also welcome and will get loved. Most of the existing checks are short Go files that took a few hours each; the code is rarely the bottleneck.

The point​

The goal isn't to be the kitchen-sink monitoring tool that does everything badly. It's to be the kitchen-sink monitoring tool that does each thing as well as the dedicated tool you'd otherwise have to install separately. That's a higher bar, I will miss it sometimes, and that's fine — that's literally what the issue tracker is for.

So: what's missing?

Tell me. I'll add it.