Skip to main content
Version: 2026.8.0

Alerting

Two kinds of "tell me when something happens" need different plumbing. Cobblr can notify you about things that happen inside a workspace. Telling you the box itself is down has to come from outside the box, because a down instance cannot send its own alarm.

Is the instance up: watch from outside​

Point an uptime monitor at the health endpoint at the address you actually browse Cobblr at:

https://<your-address>/api/v1/healthz
  • Use the browsing address: the stack binds the raw web port to loopback in every HTTPS mode, so :8088 answers another machine only on the plain-HTTP fast path.
  • Run the check from somewhere other than the Cobblr box, so a total outage still reaches you.
  • Any uptime service that fetches a URL on an interval and notifies on failure works: a hosted checker, or a small script on a second machine that curls the endpoint and posts to a webhook when it fails.
  • It returns {"ok": true, ...} while the api is serving (see Health and status).

A minimal version is a cron job on another machine:

curl -fsS --max-time 10 http://<your-box>:8088/api/v1/healthz > /dev/null \
|| curl -s -X POST -H 'Content-Type: application/json' \
-d '{"content":"Cobblr health check failed"}' \
"$DISCORD_WEBHOOK_URL"

The -f flag makes curl fail on a non-200, so an api that is down (or a box that is unreachable) triggers the notification. A Discord or Slack incoming webhook is the quick target, but any webhook URL works.

Things that happen inside a workspace​

Point a Discord or Slack incoming webhook at the matching channel in the app, then choose which events go there. Full setup is in Notifications.

  • For events Cobblr itself produces, such as an order arriving or stock dropping below its minimum.
  • Per-event routing: operational things can go to a channel and the rest can stay in the app.
  • Not an uptime check: these are for workspace activity, not for whether the server is alive, which is the outside check above.

What does not phone home​

Cobblr sends nothing on its own. There is no telemetry, no error reporting, and no update check calling out to any service. Every alert described here is one you wire up and point at a destination you control, which is also why the uptime check has to be something you set up rather than something that already exists.