Advice
Server room cooling failure: what to do when AC fails
Learn how to manage a server room cooling failure, assess temporary air conditioning, protect equipment and decide when controlled shutdown is necessary.
Reduce the heat load while restoring a route for heat to escape
When server room air conditioning fails, check temperatures at equipment air intakes, alert IT and facilities staff, reduce non-essential workloads and arrange an urgent cooling repair. Temporary air conditioning can help only if it has sufficient capacity, suitable power and a safe exhaust route. If temperatures approach equipment limits or keep rising, follow the agreed controlled shutdown procedure.
Treat a server room cooling failure as a joint building-services and IT incident. Facilities staff should handle the cooling plant, electrical supply and temporary installation; IT should prioritise services, reduce equipment demand and protect data. Nominate one incident lead so that decisions about cooling, workloads and shutdown do not conflict.
Do not assume there is a fixed safe period after cooling stops. The available time depends on equipment demand, room construction, remaining airflow and starting conditions. If there is smoke, a burning smell or water near electrical equipment, follow the site's emergency arrangements rather than continuing routine cooling checks.
Measure the air entering the servers, not just the room
Base decisions on equipment inlet temperatures and their trend, rather than a single wall thermostat. A room can feel tolerable while a rack draws in its own hot exhaust. Check the equipment manufacturer's environmental limits and your site's alarm settings; there is no universal room temperature that makes every installation safe.
Read existing rack sensors and server management alerts first. Where safe access allows, use suitable temperature probes at the front air intakes, checking different rack heights and the most heavily loaded areas. Record the time, location and reading. Compare these with previous readings to establish whether conditions are stable or deteriorating.
Also review fan-speed alerts, thermal warnings and any reported performance throttling. These can reveal local problems that a room sensor misses. Set a monitoring frequency appropriate to the rate of change, and agree who can authorise shutdown. Do not wait for automatic thermal protection to become the normal way of managing the incident.
Cut electrical demand without creating a second outage
Reducing IT power demand reduces the heat that cooling must remove. Ask the responsible IT team to pause non-essential processing, defer scheduled jobs or move suitable services to unaffected infrastructure. Prioritise changes that reduce actual equipment power rather than simply moving activity between servers in the same room.
Check dependencies before switching anything off. Storage, networking, authentication and backup systems may support services beyond the affected rack. Use the documented shutdown order and verify that any failover destination has enough computing, electrical and cooling capacity. A hurried migration can transfer the problem rather than solve it.
Use measured electrical load, where available, to assess the scale of the heat source. Most electricity consumed by IT equipment becomes heat within the cooled space. Power-distribution or UPS readings can help, but confirm what they include: a UPS may feed equipment elsewhere, and its rated capacity is not its actual operating load.
Check whether a portable unit can cover the remaining load
Select temporary cooling against the remaining heat load, not floor area alone. Server rooms concentrate electrical equipment into small spaces, so a unit suitable for an ordinary office of similar size may be inadequate. Include relevant heat from in-room power equipment, lighting, surrounding spaces and solar gain, without double-counting loads already measured.
ChillDash hires professional-grade portable air conditioners rated at 12,000–14,000 BTU/h, equivalent to approximately 3.5–4.1 kW of nominal cooling capacity. This is cooling output, not electrical consumption. If measured IT demand alone exceeds a proposed unit's nominal cooling output, that unit cannot reasonably be treated as a complete cooling replacement.
Actual usable cooling depends on operating conditions, ducting, airflow and the particular model. Ask the supplier to assess the application and confirm operating limits and condensate arrangements. Larger loads may require specialist temporary plant or further workload reduction. Multiple portable units are not automatically a workable answer: each needs suitable power, space and heat rejection.
Solve the exhaust route before accepting temporary cooling
A portable air conditioner must discharge its extracted heat somewhere that will not feed it back into the server room. Identify a suitable external exhaust route before arranging equipment. Venting into a ceiling void, enclosed corridor or neighbouring room can simply move the overheating problem and allow heat to return.
Check the permitted hose length, diameter and routing with the supplier. Keep ducting within the manufacturer's limits and avoid crushing it or improvising long extensions. A window arrangement must address sealing, weather and security. Do not prop open fire doors or alter fire-resisting construction to accommodate a hose; involve the responsible building professional.
Single-hose units exhaust room air and draw replacement air through available openings. That incoming air may be warm, humid or dusty, reducing effectiveness. Explain whether the room is sealed, windowless or served by shared ventilation when requesting advice. Where no safe exhaust route exists, specialist ducting or a different temporary cooling arrangement may be necessary.
Use a four-stage installation and commissioning check
First, confirm the room can physically accept the proposed equipment. Measure access doors, turns, available floor space and the exhaust route. Keep escape routes, rack access and electrical panels clear. Give the supplier the room layout, measured IT load, inlet readings and details of any cooling that still works.
Next, confirm electrical suitability before connection. Check the unit's supply requirements and starting characteristics against the available circuit, with an electrician where necessary. Do not assume that a spare socket means spare circuit capacity, or connect temporary cooling to a rack power strip or IT UPS without specific design approval.
Then establish where condensate will go. Some models still need drainage or tank emptying despite evaporating some moisture during operation. Keep water arrangements away from electrical equipment and walkways, and confirm who will check them. Place the unit so its cool discharge can reach equipment intakes without obstructing rack airflow.
Finally, commission the whole arrangement under the intended operating load. Check exhaust security, drainage, power supply and inlet-temperature trends across the racks. A cold stream from the unit is not proof of adequate cooling. Document alarms, inspection responsibilities and the response if the temporary unit trips or stops overnight.
Avoid fixes that circulate heat or introduce new risks
Fans alone cannot remove heat from a closed room. They may improve air distribution, but can also push hot rack exhaust towards equipment intakes. Preserve the intended front-to-back airflow where applicable, keep intake paths clear and avoid removing rack panels or rearranging equipment without understanding the existing airflow design.
Opening a door is not a dependable replacement for cooling. It helps only when cooler replacement air and a useful heat escape route exist, and may compromise security, fire separation or neighbouring spaces. Do not use evaporative coolers, ice or wet towels as improvised server-room cooling; they introduce moisture without providing controlled heat rejection.
Portable cooling may also be unsuitable where access, electrical capacity, drainage or exhaust options are inadequate. Stop repeatedly resetting a tripping unit or circuit and have the fault assessed. If safe cooling cannot keep equipment within its approved conditions, reducing service availability through a controlled shutdown is preferable to relying on uncontrolled equipment failure.
Restart in stages and keep monitoring after the repair
Restore workloads gradually only after the repaired cooling system has been checked and equipment inlet temperatures are stable. A cooler room at reduced IT demand does not prove that the system can handle normal service. Coordinate each increase in load with temperature monitoring and the engineer's commissioning advice.
Keep temporary cooling available until the permanent system has demonstrated satisfactory operation at the intended load, subject to a safe installation. Watch for local hot spots, cooling alarms and drainage problems. Record the repair findings and confirm that remote alerts reach someone who can act outside normal staffing hours.
Afterwards, update the incident plan with actual electrical demand, sensor locations, service shutdown priorities and an assessed temporary exhaust route. Resolve any uncertainty about power supplies or landlord permissions in advance. The most useful preparation is knowing which temporary arrangement can be installed safely, and when cooling limitations require IT services to stop.
Common questions
Does a UPS keep server-room cooling running?
Only if the cooling equipment is specifically connected to and supported by that system. Many IT UPS arrangements protect servers and networking, not air conditioning. During a power interruption, IT may therefore continue producing heat after cooling stops. Verify the actual electrical design rather than assuming both are protected.
Can building ventilation replace failed air conditioning?
Only where its air temperature, volume and operating arrangements can meet the load. Ordinary ventilation is not necessarily designed for concentrated IT heat. Ask the building-services engineer to assess it; do not override dampers, smoke controls or ventilation interlocks.
Should a cold-air hose be attached directly to a server rack?
Not without an approved distribution arrangement. Improvised connections can restrict the air conditioner's discharge or produce uneven rack airflow. Ask the cooling supplier and rack specialist whether ducted delivery is supported and how inlet conditions will be verified.
What information should be included in an emergency cooling enquiry?
Provide the site location, access restrictions, room layout, measured load, temperature trend, power supply and possible exhaust route. State whether continuous operation is required and who can approve installation. ChillDash delivers and collects nationwide; suitability and availability need checking, and the price is confirmed when availability is checked.
Is humidity monitoring necessary during temporary cooling?
Check humidity against the equipment manufacturer's environmental specifications, particularly if the temporary arrangement introduces outside air. Temperature alone does not describe the operating environment. Seek specialist advice if humidity is outside the permitted range or condensation appears.