A server DIMM rarely dies without warning. ECC memory fixes single-bit errors as they happen and logs them, so the log is your early warning. The skill is separating a one-off error, which is normal, from a pattern that means a module is on its way out. This guide shows what to read, what the codes mean, how to confirm the module with a swap test, and when to replace it. If the server will not start at all, see the POST troubleshooting hub. For the ECC background, see the ECC vs non-ECC guide.
The short answer
- Read the system event log and the memory error counters. They name the module and say whether the error was corrected.
- Correctable errors are the system working as designed. A repeating pattern on one DIMM is the sign to act.
- Uncorrectable errors need action. Dell's MEM0001, MEM9072 and MEM6104 all report uncorrectable events.
- Confirm with a swap test. If the error follows the module to another slot, the module is faulty.
Correctable and uncorrectable errors
Dell defines a correctable error as one the system fixed on a specific DIMM, and an uncorrectable error as one the chipset could not fix, usually with a bank number, where the affected DIMM may or may not be identifiable. The Linux kernel's EDAC documentation uses the same split: a corrected error is one ECC detected and fixed, an uncorrected error is one ECC could not fix but from which the system can usually continue, and it adds deferred and fatal classes for uncorrectable errors that are not urgent or cannot be recovered. A single corrected error is not a failure. A DIMM that logs corrected errors again and again, or any uncorrectable error, is a candidate for replacement.
What the logs and counters show
| Signal | What it means | Action |
|---|---|---|
| Correctable error on a DIMM | The system fixed an error on that DIMM (Dell); Linux EDAC calls it a corrected error | Log it and watch the count; reseat if it repeats |
| Repeated correctable errors on the same DIMM | The DIMM may be degrading | Swap test; plan a replacement |
| MEM9072 | Patrol scrub found an uncorrectable error that was not consumed; usually no impact unless the OS uses that memory | Act now: reseat and swap test |
| MEM0001 | An uncorrectable event was consumed; may cause a reboot | Replace after a swap test confirms |
| Uncorrectable (UE) in Linux EDAC | ECC could not fix the fault; the system can usually continue | Treat as urgent |
| Errors on several DIMMs | Failed memory, or a board or processor fault | Contact the server maker; collect a diagnostic report |
On Linux, EDAC exposes counters under /sys/devices/system/edac/, with a corrected error count for each rank, and reports errors with the layer positions of the affected hardware, such as channel and slot. Dell servers log the event in the System Event Log in iDRAC with the slot. Dell adds that these codes are examples and servers can report other memory events, so read the full message for your model.
Dell's order of actions
- Check current firmware. Update the CPLD, iDRAC and BIOS if they are not current.
- Power off fully and reseat the modules. A loose module produces errors that look like a failing one.
- Clear the System Event Log. New errors are then easy to read.
- Reboot and collect a diagnostic report (SupportAssist TSR on Dell) and check for new memory messages.
- For one DIMM with errors, follow the Lifecycle Controller log. For several DIMMs, contact the server maker.
Confirm the module with a swap test
Move the suspect module to a different slot, ideally on the other processor, and watch where the error is reported. Dell's conclusions are: if the error disappears, reseating fixed it; if it follows the module, the module has most likely failed; if it stays with the socket, the board or processor is the suspect, and a processor swap decides which. The full method and results table are in RAM or motherboard.
When you can take the server down: test it
A memory test finds faults the logs have not caught yet. MemTest86 suggests removing modules to isolate a failing one, repeating passes to check that errors are reproducible, and notes that not every error is bad RAM, since the test also exercises the processor, caches and motherboard. It also lists marginal conditions, temperature sensitivity and power supply voltage ripple. If errors only appear with modules together, check the layout and the mix; see mixing RAM modules.
Replacing the module
Replace with the same type, generation, voltage and capacity as the rest of the set, so the server's rules still hold. If the failing module is one of a set, replacing only the faulty one is normal. Check the part first with how to identify an incompatible RAM module. Browse DDR4 RDIMM and DDR5 RDIMM, or send the server model and the slot from the log through the quote form. For a server that reports less memory than it holds, see why a server does not recognize all installed RAM.
Frequently asked questions
How do I identify a failing server DIMM?
Read the system event log and error counters for the module and the error type. Repeated correctable errors on one DIMM, or any uncorrectable error, point to a failing module. Confirm with a swap test.
What is the difference between correctable and uncorrectable memory errors?
A correctable error is one ECC fixed. An uncorrectable error is one it could not fix. Dell and the Linux EDAC documentation use this split, and a repeating pattern of correctable errors or any uncorrectable error needs attention.
What do Dell memory codes MEM0001 and MEM9072 mean?
Dell lists MEM0001 as an uncorrectable event that was consumed and may cause a reboot, and MEM9072 as an uncorrectable error found by patrol scrub that was not consumed, usually with no impact unless the OS uses that memory.
Should I replace a DIMM after one correctable error?
Usually not. A single corrected error is the ECC working. Reseat the module, watch the count, and replace it if errors repeat on the same DIMM.
How do I confirm a DIMM is bad?
Move it to another slot and see whether the error follows. If it follows the module, the module most likely failed, according to Dell's swap-testing conclusions.
Sources
Read on October 30, 2026: Dell PowerEdge memory issues troubleshooting guidelines, Dell article on troubleshooting memory errors by swap testing, Linux kernel EDAC documentation and MemTest86 troubleshooting guide. 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, Linux and MemTest86 are trademarks of their respective owners, used here only to identify sources and compatible systems.