SRE Daily

A 30-minute weekday meeting for decisions, not status: what happened, are we healthy, what's stuck, what's risky today, and who needs help.

Purpose

Start every day with the same shared picture: what happened, are we healthy, what's stuck, what's risky today, and who's on interrupts. It's for decisions and unblocking, not a round-robin status read-out.

Who's in the Room

The SRE team, including whoever is on call.
The NOC shift lead, for minutes 08–12, if time zones allow. If they can't attend, use their written handoff.
A facilitator who rotates weekly. You don't have to run it every day, and having others facilitate builds the team and frees you to listen.
You: listen for what's stuck, take the blockers, and protect focus time.

Before the Meeting

On-call posts three lines in the team channel 10 minutes before: what paged, what's still open, and anything weird.
One person checks the deploy and change calendar for risky changes today. This rotates with the facilitator.
You: read the on-call note and the NOC handoff.

The Agenda

Time	Block	Ask
00–05	Last 24 hoursIncidents, pages, and near misses, from on-call. Only what matters.	"What happened that we'd regret not talking about?" Then: "Has it happened before?"
05–08	Health verdictGreen, yellow, or red, plus one sentence of evidence.	"Green, yellow, or red, and why?"
08–12	NOC handoffOpen items from the overnight shift, each one getting an SRE owner. Thank the NOC for anything they caught well.	"What did the NOC hand over, and is anything still open?"
12–20	Active & stuckOpen incidents, escalations to pods, and anything blocked more than two days.	"What's stuck, and who or what is it waiting on?" Blocked more than two days: "That one's mine."
20–25	Today's risk & interruptsDeploys, launches, provider maintenance, and config changes. Name today's interrupt handler.	"What's changing today that could hurt us?" and "Who's on interrupts?"
25–30	Asks & recapWho needs help, the parking lot assignments, and actions read back.	"Is anyone carrying too much right now?"

Calendar invite text: SRE Daily

SRE Daily: weekdays, 30 min (ends early if we're done)
      Purpose: a shared picture, unblock what's stuck, and protect focus time. Not a status read-out.

      00-05  Last 24 hours: incidents, pages, near misses (on-call)
      05-08  Health verdict: green / yellow / red + one sentence why
      08-12  NOC handoff: open items from overnight, each gets an SRE owner
      12-20  Active & stuck: open incidents, pod escalations, anything blocked > 2 days
      20-25  Today's risk: deploys, launches, provider maintenance. Name today's interrupt handler.
      25-30  Asks & recap: who needs help, parking lot, actions read back

      Prep: on-call posts 3 lines in the channel 10 min before (what paged / still open / anything weird).
      Rules: over 2 minutes = parking lot. No live debugging in the meeting. Facilitator rotates weekly.

Questions to Ask

Bring three. Each has a follow-up for a weak answer, and coaching notes on what good sounds like.

Example in Practice

Minute 6, health verdict. The facilitator asks, "Green, yellow, or red?" On-call says "green." You: "What would make it yellow?" On-call pauses: "Actually, the slowest 5% of answers has been creeping up since 6 a.m." Now it's yellow, with a reason, before any customer notices.

Minute 9, NOC handoff. The NOC flagged "assistant slow for some users, 02:00–02:40, resolved itself." Nobody on SRE had seen it. You: "That lines up with the creep. Sam, can you take both and report back at tomorrow's daily?" Then, to the NOC shift lead: "Good catch, thank you for writing it up." In other words, the NOC's note turned into an early warning, and the NOC heard that their work matters.

What Actually Happens

It becomes a round-robin where everyone reads their ticket list.: Counter: kill the round-robin. Go by agenda block, not by person. If someone has nothing stuck and nothing risky, they say nothing, and that's a good daily.
Two engineers start debugging live while everyone else watches.: Counter: "Great, that's a parking lot item. You two take it right after." Say it kindly and every time, until the team says it before you do.
People stop coming because it always runs over.: Counter: hard stop at 30, end early when you can. A daily that regularly ends at 18 minutes earns its attendance.
The NOC handoff gets skipped when the meeting runs long.: Counter: it's minutes 08–12 on purpose, early, before the long discussions. Skipping it tells the offshore team their work doesn't matter.
Nobody admits to being overloaded in front of peers.: Counter: ask in the room anyway, so it's normal. Then follow up privately with whoever had the worst night.

Rules for Every Recurring Meeting

Same format every time. People prepare for a format they can predict.
Hard stop at 30 minutes. End early if you're done. A meeting that respects time gets attended.
Anything over 2 minutes goes to the parking lot, with two named people to take it right after.
No action without an owner, a date, and what "done" means.
A three-line written recap within the hour. It reaches people who missed it, and it becomes the record.