For · Ops
The on call rotation, designed in.
Quiet hours per channel. Critical bypass that ignores quiet hours during real outages. Recovery emails that go out the moment the response code clears. Reports that read like a calm engineer wrote them.
Quiet hours
The on call channel sleeps. The on call still gets the page.
Set the on call channel to critical only between 22:00 and 07:00 in the rotation timezone. Outage and certificate expiry pages still fire. The four times daily status digest stays silent.
The page that wakes you up is the one that should.
Critical bypass
Quiet hours respect outages
Acknowledgement
Who is on it. When. With which note.
Acknowledgement records the user, the timestamp, and an optional note. The note shows up in the next status digest so the rest of the team can see what is happening without asking.
Acknowledgement does not close the incident. Resolution does. The two are independent because they answer two different questions.
engager.rookhq.com / @realm / alerts
api.rookhq.com · Response time
criticalopenP99 spiked to 8.4s for 3 consecutive pings
7m ago · Delivered to Ops bot · On call
rookhq.com · SSL certificate
warningopenCertificate expires in 17 days (warn threshold = 30d)
23h ago · Delivered to Ops bot · ops@team.rookhq.com
@hlotech · GitHub PAT
warningacknowledgedPAT scopes missing repo:read · 4 org repos not enumerable
3h ago · Ack'd by Aravindh · Delivered to GitHub admin chat
Things ops people care about
Six things that matter at 3am.
Page only on signal
The cold start guard, the consecutive failures rule, the sustained anomaly count. Three filters before a single buzz.
Active days
Saturdays off for the marketing channel, always on for the on call. Per channel.
Critical bypass
Quiet hours always honour critical kinds. Outages, expiries, watchdog failures wake you up.
Rotation handoff
Mute window per channel during a vendor migration or a controlled outage. Auto resume on the date you pick.
Notes in the digest
The acknowledgement note from the incident appears inline in the next status report. Compounding documentation.
Replay a missed page
Telegram dropped the message? Replay to the channel without resending the rest.
restrained voice everywhere
Sentence case. No exclamation marks. Restrained colour. Reports look like internal documents, not marketing.
Dashboard at 3am
Solid backgrounds, hairline borders, no glassmorphism. Readable with squinted eyes on a phone.
Audit forever
Every action is recorded with actor and timestamp. Postmortem prep is a SQL query, not a Slack archaeology session.