How can we watch them?

Servers fail at three in the morning. Watching is being told the right thing, at the right time.

November 2026Used in ZilaOS, Carabdheer, ZilaFSM
A night street from above, one lit window

A server runs out of memory at three in the morning. The app hangs. Nobody knows until a customer calls at nine. Every team running its own servers knows this story. Most of us have lived it.

Carabdheer is Somali for “long tongue”: the one who tells. That is the whole job.

One place, every app

Each tool we tried watched one app, or lived in someone else’s cloud. So we built one place for every app on every server. Errors grouped into issues. One log stream with a live tail. Hosts and containers with the reason they restarted. Failed jobs, backups, mail, certificates.

  1. 03:02hostdb-2 memory 86% for 4 min
  2. 03:04web-1QueryException: deadlock found → issue #214
  3. 03:04agentworker restarted · exit 137 · out of memory
  4. 03:10cronbackup nightly ok · 2.1 GB
  5. 07:30summarymorning summary sent · 1 issue, 1 restart
Example log lines from one night.

We watch the machines so people can sleep.

Tell, don’t shout

An alert that fires all night gets muted by morning. So Carabdheer has quiet hours and a morning summary, and it speaks on the channels people actually read: email, Telegram, WhatsApp.

Your data stays yours. Each workspace has its own database. Old detail fades on schedule. The daily summaries stay forever, so you can always see how a month really went.

If you build, dream, or make, you belong here.

Find your way in