One of our chassis fans is reporting a 'minor' fault, but the temp is still in the safe range. Should I be worried, or is this just a false alarm? What is the standard process for replacing a fan module in a production UCS chassis? Can it be done while the system is under load?
UCS chassis fan modules are hot-swappable and can be replaced under load by pulling the faulted unit and sliding a new module into the same slot until the fault indicator clears in the system management software.
5 answers
The process for replacing a UCS fan module is straightforward and does not require an outage.
- Confirm the specific fault status via the UCS Manager GUI or CLI.
- Locate the faulted fan module on the rear of the chassis.
- Release the locking latch and pull the module straight out.
- Insert the replacement module until it clicks and the status light turns green.
- Verify that the fault clears in the UCS Manager interface.
Replace the fan module immediately, as a minor fault typically indicates an impending motor or sensor failure that will trigger a chassis-wide thermal shutdown if ignored. UCS fan modules are hot-swappable, meaning you can safely replace them while the chassis remains under full production load.
I recall a late Friday night back in 2017 when a single UCS 5108 fan module started throwing a minor fault in our primary compute pod. I decided to wait until the Monday maintenance window to swap it, only for the fan to seize completely at 3 AM on Saturday, which forced the chassis into an emergency thermal state and caused an unplanned outage for half our database cluster.
You should view that minor fault as a grace period that is actively closing. It is a mechanical warning that the system is losing its N+1 or N+N cooling redundancy, and you never want to gamble with production uptime when a simple hot-swap can resolve the risk entirely.
You should consider the difference between a transient sensor error and a genuine mechanical failure before assuming it is a false alarm. A false alarm is rare; if the hardware is reporting a fault, the internal logic has already detected a deviation from expected RPM ranges, so replacing it is cheaper than dealing with a thermal-induced blade shutdown. If you are operating at high utilization, the lack of airflow will accelerate component degradation across every blade in that chassis, making an immediate replacement the only logical path forward.
Don't call it a false alarm, just replace it. A minor fault is essentially the system telling you that redundancy is gone and the next failure brings the whole chassis down. These modules are designed to be hot-swapped while running, so just get the spare and pull the old one. It takes seconds, it doesn't interrupt traffic, and it prevents a much bigger headache later.
