Servers run for years, and the log fills with memory messages long before anything fails. The hard part is deciding when a message means a module must go. Replace too early and you pay for parts and downtime you did not need. Replace too late and you take an unplanned outage. This guide gives clear criteria based on Dell's published guidance, a decision table, and the steps to run before you order. For identifying the failing module, see how to identify a failing server DIMM.
The short answer
- Replace a module when a test shows it is the cause. An error that follows the module to another slot means the module has most likely failed, per Dell's swap-test conclusions.
- Do not replace for correctable errors alone. Dell says replacement is recommended only when self-healing fails on its 14G and 15G DDR4 servers.
- Treat uncorrectable errors as urgent, and confirm the module before ordering.
- Do the free steps first: update firmware, power off, reseat, clear the log and observe.
The decision table
| What you see | Replace? | Why |
|---|---|---|
| One-off correctable errors | No | Dell says correctable errors alone do not warrant replacement on its 14G and 15G self-healing platforms |
| Correctable error threshold event (MEM0802, MEM5104) | Not yet | The BIOS schedules self-healing at the next reboot; reboot at your normal maintenance window |
| Self-healing fails (MEM0805, MEM7114) | Yes | Dell recommends physically replacing the DIMM |
| Uncorrectable error on a DIMM, confirmed by swap test | Yes | The error follows the module, so the module has most likely failed |
| Uncorrectable errors on several DIMMs | Not first | Dell's guidance for multiple DIMM errors is to contact Dell support |
| Errors stay with the slot after the module moves | No, not the module | The board or processor is the likely cause |
| Planned capacity or speed upgrade | Your call | A refresh decision, not a fault; see the refresh guide |
What Dell says about correctable errors
On DDR4 RDIMMs and LRDIMMs in Dell 14G and 15G Intel and 15G AMD PowerEdge servers, a correctable error threshold event (MEM0802 or MEM5104) schedules self-healing. The BIOS uses spare rows in the DRAM to permanently replace a bad row, and it runs at the next reboot even when correctable error logging is disabled. If self-healing fails, the system logs MEM0805 or MEM7114 and recommends replacing the DIMM. Dell's article says correctable errors alone do not warrant replacement. These figures apply to the platforms the article names; other servers handle correctable errors differently, so check your model.
What Dell says about uncorrectable errors
Dell's memory troubleshooting guidelines list MEM0001 as an uncorrectable event consumed, which may cause a reboot, MEM9072 as a patrol scrub uncorrectable error that was not consumed and usually has no impact unless the OS uses that memory, and MEM6104 as an uncorrectable error. Dell's recommended order is to check firmware versions, update the CPLD, iDRAC and BIOS, power off fully, reseat the modules, clear the System Event Log, reboot, and collect a SupportAssist report. For one DIMM with errors, Dell points to the Lifecycle Controller log for the recommended action. For several DIMMs, contact Dell.
Before you order: confirm the module
- Record the slot from the log.
- Update firmware and reseat, then clear the log and watch.
- Swap the module to another slot, ideally on the other processor, and see where the error is reported. The method is in RAM problem or motherboard problem.
- If the error follows the module, order a replacement. If it stays with the slot, do not.
The cost of waiting
This part is our judgment, not a vendor rule. If a server runs a workload where an unplanned reboot is expensive, replacing a module after a confirmed uncorrectable error at your next maintenance window is cheaper than waiting for the next one. If the server is spare capacity, you can wait for more data. Keep a spare module of the same type on the shelf for critical systems, so replacement is a swap rather than a purchase order.
Choosing the replacement
Match the failing module on type, generation, voltage, speed, capacity and rank, so the population rules still hold. See how to identify an incompatible RAM module and mixing RAM modules. If you are replacing for capacity or speed rather than a fault, the data center memory refresh guide covers planning. Browse DDR4 RDIMM and DDR5 RDIMM. Modules supplied by NEMIX RAM carry a lifetime warranty; register your purchase to activate it.
Frequently asked questions
When should you replace a server memory module?
Replace it when a test shows it is the cause: an uncorrectable error that follows the module to another slot, or failed self-healing on platforms that report it. Correctable errors alone do not warrant replacement.
Do correctable memory errors mean I need to replace the DIMM?
Not by themselves. Dell says correctable errors alone do not warrant replacement on its 14G and 15G self-healing servers. Replace when self-healing fails or errors turn uncorrectable.
What does it mean when self-healing fails on a Dell server?
The system logs MEM0805 or MEM7114 and Dell recommends physically replacing the DIMM.
How do I confirm the memory module is the problem before buying?
Update firmware, reseat, clear the log, then swap the module to a different slot. If the error follows the module, the module has most likely failed.
Should I keep spare server memory?
For critical servers, yes. A spare of the same type lets you replace a confirmed bad module at the next maintenance window instead of waiting for a purchase.
Sources
Read on October 30, 2026: Dell article on managing correctable error threshold events, Dell PowerEdge memory issues troubleshooting guidelines and Dell article on troubleshooting memory errors by swap testing. Rules differ by model and generation, so confirm your exact system in its own manual.
Memory matched to your systems
NEMIX RAM is a US-based memory specialist, in business since 1993. Every module is researched, matched, and compatibility-tested by NEMIX RAM, is JEDEC-compliant, and is backed by a lifetime warranty; register your purchase to activate it. Purchase orders, net terms and volume pricing are available for IT teams, resellers, and government and education buyers. Request a quote with your system models, processors and quantities, or contact our sales team. Browse DDR4 RDIMM, DDR5 RDIMM and LRDIMM, or see our data center server memory page.
NEMIX RAM is an independent memory supplier and is not affiliated with the companies named here. Dell, PowerEdge, iDRAC, Intel and AMD are trademarks of their respective owners, used here only to identify sources and compatible systems.