[–] [S] -2 points 7 hours ago

yeah, the matrix room approach is clean. that's basically the notification hub I'm talking about, you just built it yourself with synapse + bridges. My selfhosted setup is quite simple and small, I would prefer avoid any overkill setup, that's all ^^

and no argument on the fixing part. I'm not trying to get anyone debugging from a phone, that's madness. logs on a real keyboard, every time.

the only bit I actually care about is the "fix now or leave it til morning" call when I'm away from home, plus enough context in the alert to make that call. sounds like yours already handles that fine.

so fair enough, if you've got synapse bridging everything you're not the one who'd need this. realistically the target is people who'd never stand up a matrix server just for alerts. anyway thanks for spelling it out, that's useful.

  • source
  • parent
  • context
  • [+] [S] -12 points 7 hours ago (1 child)

    Yeah, that's fair — if you've got Zabbix/Nagios wired to alerts + automation that acts on them, you're honestly not who this is for. You've already solved it the proper way.

    The one thing I'd push back on: I'm not really pitching a dashboard you stare at. It's push-first too — the "board" is just what you glance at the one time you actually open it after a buzz. The gap I keep hitting isn't "I want to watch graphs", it's "alert fires, I'm on my phone away from home, and I want to see what else is affected and tap straight into it" without spinning up a VPN and digging for a bookmark.

    But your setup is basically the endgame, so point taken — for people who've built that pipeline, this is redundant. That's exactly the line I'm trying to find, so thanks.

  • source
  •  

    Hey folks,

    Quick gut-check before I sink weeks into building the wrong thing.

    The itch: I self-host the usual suspects — a Storj node, Pi-hole, WireGuard, Uptime Kuma, a blog, the lot. Keeping an eye on all of it from my phone is a mess. Four browser tabs, four logins, reachable only on the LAN, and nothing buzzes me when a service quietly dies at 3am. Every "nice" mobile monitoring app I tried wants a cloud account and phones home — the exact opposite of why I self-host in the first place.

    So I want to build the app I actually want to use: one board for the whole stack, push alerts that deep-link straight to whatever went down, a home-screen widget, direct talk to the APIs you already run (Uptime Kuma, Pi-hole, Portainer, your node…), over the LAN or your own VPN — no middleman server, no account, no telemetry, open-core.

    Before writing a line of app code, I'm validating demand. And I'll put my money where my mouth is: the landing page and the waitlist both run on a Debian box in my living room. The form that takes your email is a ~150-line stdlib Python service behind Caddy, writing to a flat file on my own disk — no Google Form, no SaaS, no third party sees a single byte. If I can't self-host a waitlist, I've no business building this.

    👉 https://selfhold.duckdns.org/

    Honestly I'd rather have your take than your email: would you use it, or is your setup fine? What annoys you most about keeping tabs on your stack? (On the page) would you pay anything, or is free-only the only version that makes sense?

    And yes — I know Uptime Kuma / Beszel / Gatus exist and are great. This isn't trying to replace your monitoring server; it's the mobile, unified, privacy-first front-end I keep wishing those had. If that's pointless, tell me — rather hear it now than after building it.

    (It doesn't exist yet. This is me deciding whether to build it. No hype, no VC — just conviction and a mini-PC in the Jura.)