Skip to content
Powering what’s next.From single upgrades to datacentre builds.Wholesale Servers
Basket 0
Browse categories
Build a server

14 November 2026 · Admin

How to Replace a Failed PSU in HPE and Dell Servers

Need to replace failed PSU in an HPE or Dell server? Identify the fault, confirm part compatibility and restore redundant power with less downtime, safely.
Share this article
WhatsApp LinkedIn Facebook X Email

A power-supply alert is not automatically a power-supply failure. On a production HPE ProLiant or Dell PowerEdge server, the quickest way to replace failed PSU hardware without creating a larger outage is to establish whether the issue is the module, the mains feed, the power distribution unit (PDU), or a reporting fault. A failed PSU in a correctly configured redundant pair is usually a planned maintenance task. In a non-redundant configuration, it is an immediate availability risk.

The practical objective is simple: retain stable power to the server, fit a compatible replacement, and return the system to its intended redundancy level. That requires more than matching a wattage figure on a label.

Confirm the PSU has actually failed

Start with the server management logs and the physical status indicators. HPE iLO and Dell iDRAC normally record power-supply events, input loss, predictive faults and redundancy state. Compare the reported slot number with the LED on the rear of the chassis. A PSU showing no input LED may have lost its AC supply rather than suffered an internal failure.

Check the basics before ordering hardware. Verify that the IEC lead is seated, the PDU outlet is live, and the circuit has not tripped. If the platform uses two independent feeds, confirm the other PSU is connected to a separate PDU or UPS as designed. Moving the suspect module's power lead to a known-good feed can help distinguish an upstream issue from a failed unit, provided site change controls permit it.

A green output LED does not always clear a PSU. Intermittent faults, fan errors and voltage regulation warnings may still be logged by iLO or iDRAC. Conversely, a failed status can remain after a transient input event until the management controller refreshes. Record the exact alert text, chassis model, service tag or serial number, and fitted PSU part number before removing anything.

Identify the correct replacement PSU

Enterprise server PSUs are not generic ATX components. Mechanical format, connector arrangement, firmware support, airflow design, input rating and server generation all matter. A 750W HPE Common Slot unit may physically resemble another Common Slot PSU while carrying a different option number, efficiency class or supported platform list. Dell hot-plug PSUs likewise vary across PowerEdge generations and chassis families.

Use the original module label as the primary reference. Capture the manufacturer spare number, option number and wattage, then cross-check it against the server model and generation. For example, an HPE DL380 Gen9, DL380 Gen10 and DL380 Gen10 Plus should not be treated as interchangeable PSU estates simply because their chassis layouts look familiar. The same principle applies across Dell Gen12, Gen13 and Gen14 systems.

Where the server contains two PSUs, matching the existing unit is normally the lowest-risk route. Mixed wattages can reduce available redundancy, generate management alerts or restrict the server's supported power budget. Mixing efficiency ratings, such as Platinum and Titanium, is also best avoided unless the OEM documentation explicitly permits the pairing for that platform.

The required wattage depends on the actual configuration, not the base server specification. A lightly configured single-CPU system may operate well below its PSU capacity, while dual processors, high-density memory, multiple drive cages, GPU accelerators and expansion cards materially increase load. The surviving PSU must be capable of supporting the full current draw by itself if you expect N+1 redundancy.

For refurbished hardware, inspect the replacement before installation. Check the pull handle, release latch, edge connector, fan grille and label condition. Confirm that no pins are damaged and that the unit is free from dust build-up or signs of overheating. KahnServers customers sourcing a replacement for an established HPE or Dell estate should purchase against the exact server and part number rather than wattage alone.

Plan the change around the power architecture

If one PSU remains healthy and the server reports that redundancy is degraded rather than lost, most rack servers support hot replacement. Leave the server powered on, retain the working PSU and replace only the failed module. This avoids an unnecessary service interruption.

Hot-swap capability is conditional. Do not remove a second PSU if the first is already reporting warnings, if the server is on a single feed with an unstable PDU, or if the workload exceeds the capacity of one module. A server may continue running on one PSU but no longer have protection against a further power event.

A planned shutdown is safer where the affected server has a single PSU, both modules are suspect, or the hardware is operating close to the remaining PSU's limit. Shut down guest workloads and applications in the correct order, then power off through the operating system before disconnecting mains power. Do not pull a PSU from a powered chassis merely to inspect it if the platform is not designed for hot-plug operation.

How to replace a failed PSU safely

For a hot-swappable, redundant configuration, the replacement process is straightforward but should be controlled:

  • Confirm the healthy PSU is online and the server has sufficient remaining power capacity.
  • Identify the failed bay using the management alert and PSU LEDs, not position alone.
  • Disconnect the AC lead from the failed module and release its retaining latch.
  • Withdraw the module using the handle, keeping it level to avoid stressing the connector.
  • Slide the replacement fully into the empty bay until the latch engages, then reconnect its AC feed.
  • Allow the PSU to initialise and verify input, output and redundancy status through the chassis LEDs and management controller.
Use the same feed arrangement as the original design. In a dual-feed deployment, PSU 1 and PSU 2 should normally return to separate protected power paths. Connecting both PSUs to the same PDU outlet bank may keep the server running, but it removes the resilience that redundant hardware was intended to provide.

Do not open the failed module. Server PSUs contain capacitors and are field-replaceable units, not field-repairable items. The correct action is to replace the module and retain the failed part for warranty, return or recycling procedures as applicable.

Validate the server after replacement

A green LED is the first check, not the final check. Review iLO or iDRAC after the replacement and confirm that both PSUs are present, healthy and receiving input power. The system should report power redundancy as enabled or redundant, subject to the platform's configured policy and available capacity.

Clear or acknowledge historical alerts only after confirming that no current fault remains. Check event logs for repeated input-loss messages, thermal warnings or fan alerts. A new PSU that immediately reports an issue may point to a faulty AC lead, an overloaded PDU circuit, incompatible hardware or a chassis backplane problem rather than a second failed module.

It is also sensible to review power consumption after a change, particularly on virtualisation hosts and storage-heavy servers. If sustained draw is close to the capacity of one PSU, the pair may not provide the redundancy level assumed by the configuration. The answer may be higher-wattage compatible units, reduced load, or a revised power design rather than another like-for-like replacement.

When a PSU failure points to a wider issue

One failed power supply after years of service is usually an isolated component event. Multiple PSU faults, repeated loss-of-input alerts or failures across several servers deserve wider investigation. Look for high rack temperatures, blocked airflow, contaminated environments, UPS transfer events, failed PDU outlets and poor cable retention.

Age also matters in refurbished server estates. Proven platforms can remain productive for many years, but PSU fans and internal components are still subject to service life. Holding a tested compatible spare for critical HPE ProLiant and Dell PowerEdge systems can be more cost-effective than waiting for a failure to become an urgent procurement exercise.

Restoring redundancy is the real finish line. Once the replacement is recognised, correctly fed and reporting healthy alongside its partner, document the part number and incident details. That record makes the next maintenance decision faster and helps keep an otherwise dependable server in service.

Share this article
WhatsApp LinkedIn Facebook X Email
Back to Blog