Data center shift handover checklist for the next shift

The handover is where the hall changes hands

At 06:00 or 18:00 a data center changes hands. The outgoing lead knows which cooling unit has been running hot, which change window slipped, and which alarm is noise. The incoming lead knows none of it, and has ten minutes to learn it.

Most handovers are a verbal walk-through plus a log. That works on a quiet night. It fails on the night it matters, and it fails in predictable ways.

Why shift handovers fail

• Status without evidence. “Cooling is fine” is an opinion. The incoming lead cannot tell whether it means “I checked the readings an hour ago” or “nothing has alarmed”.
• Open items live in one person’s head. A deferred filter change, a verbal approval, a half-finished MOP step. If it is not written down, it leaves the building with the outgoing lead.
• Alarm noise. A list of forty acknowledged alarms hides the two that need action. The next shift needs to know what is still active and what actually changed.
• Unknowns presented as fine. A generator load test that nobody can find the date for does not get mentioned, because nothing visibly wrong comes with it. That is the most dangerous line item on the page.

A good template fixes these four things: every status carries its evidence, every open item is written down, alarms are summarized by state, and unknowns are named out loud.

A practical handover template

Use this as a starting point and adapt it to your hall. Keep each line short, and attach the evidence (a reading, a work order number, a time) rather than a conclusion.

1. Hall or room summary

• Hall or room name, shift date, outgoing and incoming lead.
• Overall state in one line, using the ratings below.
• The one or two things the incoming lead must know first.
• Racks or rows with a known issue, and the reason.

2. Cooling

• Units in service, units on standby, units out of service, and why.
• Any unit running on redundancy, and what that does to your margin.
• Filter and service intervals: last service date and next due date for each unit.
• Supply and return temperature readings, with the time they were taken.
• Any hot spots found on the last walk-round.

3. Power

• UPS state: online, on battery or in bypass, and the last battery test date.
• Generator: last load test date and result, fuel level, any open defects.
• ATS: state and last transfer test date.
• A and B feeds: both live, both carrying load, any rack on a single feed.
• Any PDU or breaker near its limit.

4. Open work and change windows

• MOPs in progress: which step you are on, what is done, what is next.
• Planned windows in the next shift, with owner and approval.
• Deferrals: what was postponed, who approved it, and until when.
• Work that started and was left in a partial state.

5. Alarms

• Active alarms: what, since when, and what has been done.
• Acknowledged but not resolved: who owns each one.
• Cleared since last shift, where the pattern matters (repeats, flapping).
• Alarms known to be false, with the reason, so the next shift does not chase them.

6. Access and safety

• Contractors or visitors on site or expected, and their scope.
• Doors, cages or zones with changed access.
• Active permits, lockouts or isolations.
• Any safety event or near miss since last shift.

7. Unknowns to verify first

• A plain list of things nobody on the outgoing shift could confirm.
• For each one: what is missing (a reading, a date, a sign-off) and who can supply it.
• The incoming lead reads this section first, not last.

8. Sign-off

• Outgoing lead: “I have handed over everything I know.”
• Incoming lead: “I have read it and I accept the hall in this state.”
• Time of sign-off. Two names, not one.

How to rate each item

Rate every line with one of four states. The ratings are only useful if the rules are the same on every shift.

• Ready: the requirement is met and you can point to the evidence. A reading, a completed work order, a dated test.
• Partially ready: the requirement is met in part, or met with reduced margin. The reason is stated.
• Not ready: the requirement is not met. The reason and the owner are stated.
• Unknown: nobody can show the evidence either way.

Why unknown is never a pass

Unknown is not a soft form of ready. It means the evidence is missing: the record was not entered, the reading is stale, the test date cannot be found. If unknown passes, every gap in your records quietly becomes a clean bill of health, and the handover reports confidence the hall has not earned.

Unknown is also not the same as not ready. Treating every missing record as a failure floods the next shift with false alarms. The right move is to name it, give it an owner, and make it the first thing the incoming lead verifies. Unknowns get checked. They do not get a green tick.

How N+1 changes the rating

Take a cooling hall designed N+1 with four units, where three carry the load. One unit goes down with a fan fault.

• Partially ready: the three remaining units carry the load with temperatures in range, and the failed unit has a work order. The hall still meets its requirement, but the margin is gone. The next shift needs to know that a second failure is now an incident.
• Not ready: the remaining units cannot hold temperature, or a second unit is already out, or the failed unit’s repair has no owner and no date.

The difference is margin. One unit down inside the redundancy you designed for is a degraded hall, not a failed one. Rating it “not ready” teaches people to ignore the rating. Rating it “ready” hides the lost margin. “Partially ready, with the reason” says what is true. Write the rule down for each system (cooling, UPS, feeds) so two different leads reach the same rating.

Make the handover read, not rebuilt

A template on a page beats memory, but it still gets rebuilt by hand every shift: someone gathers the readings, checks the work orders, and types the summary. The effort goes into assembling the picture instead of acting on it.

SuperAssets keeps the standard for each type of equipment, so “ready” means the same thing on every shift. Every rack, unit and hall gets a verdict with its reason, built from the records your team keeps. Missing evidence shows as unknown, never as a pass, and a hall shows how many of its racks are unknown so the incoming lead knows what to verify first. Roll-up rules let a hall show partially ready, with its reason, when one of its units is down instead of failing outright. The handover is read, not rebuilt.

SuperAssets does not replace your BMS, DCIM or monitoring. Alarms stay where they are, and the walk-round stays. It tells the walk-round where to look first. A ready verdict means the requirements you defined were met by the records available. It is not a guarantee of uptime.

• See how it works for halls, racks, cooling and power on the data center page.
• Open a hall in the readiness explorer and see a verdict with its reasons.
• Start with a free readiness snapshot: pick one rack row, send what you already have, and we walk you through the result live.