Skip to main content

Memory problems

Use this information to resolve issues related to memory.

Common memory problems​

DCPMM specific problems​

Displayed system memory is less than installed physical memory​

Complete the following steps until the problem is solved:
Note
Each time you install or remove a memory module, you must disconnect the server from the power source; then, wait 10 seconds before restarting the server.
  1. Make sure that:

    • No error LEDs are lit on the operator information panel.

    • No memory module error LEDs are lit on the system board.

    • Memory mirrored channel does not account for the discrepancy.

    • The memory modules are seated correctly.

    • You have installed the correct type of memory module (see Specifications or DC Persistent Memory Module (DCPMM) setup for requirements).

    • After changing or replacing a memory module, memory configuration is updated accordingly in the Setup Utility.

    • All banks of memory are enabled. The server might have automatically disabled a memory bank when it detected a problem, or a memory bank might have been manually disabled.

    • There is no memory mismatch when the server is at the minimum memory configuration.

    • When DCPMMs are installed:

      1. If the memory is set in App Direct or Mixed Memory Mode, all the saved data have been backed up, and created namespaces are deleted before any DCPMM is replaced.

      2. Refer to DC Persistent Memory Module (DCPMM) setup and see if the displayed memory fits the mode description.

      3. If DCPMMs are recently set in Memory Mode, turn it back to App Direct Mode and examine if there is namespace that has not been deleted (see DC Persistent Memory Module (DCPMM) setup).

      4. Go to the Setup Utility, select System Configuration and Boot Management > Intel Optane DCPMMs > Security, and make sure all the DCPMM units are unlocked.

  2. Reseat the memory modules, and then restart the server.

  3. Check the POST error log:

    • If a memory module was disabled by a systems-management interrupt (SMI), replace the memory module.

    • If a memory module was disabled by the user or by POST, reseat the memory module; then, run the Setup utility and enable the memory module.

  4. Run memory diagnostics. When you start a solution and press the key according to the on-screen instructions, the LXPM interface is displayed by default. (For more information, see the “Startup” section in the LXPM documentation compatible with your server at Lenovo XClarity Provisioning Manager portal page.) You can perform memory diagnostics with this interface. From the Diagnostic page, go to Run Diagnostic > Memory test or DCPMM test.

    Note
    When DCPMMs are installed, run diagnostics based on the mode that is set presently:

    • App Direct Mode:

      • Run Memory Test for DRAM memory modules.

      • Run DCPMM Test for DCPMMs.

    • Memory and Mixed Memory Mode:

      Run both Memory Test and DCPMM Test for DCPMMs.

  5. Reverse the modules between the channels (of the same processor), and then restart the server. If the problem is related to a memory module, replace the failing memory module.
    Note
    When DCPMMs are installed, only adopt this method in Memory Mode.

  6. Re-enable all memory modules using the Setup utility, and then restart the server.

  7. (Trained technician only) Install the failing memory module into a memory module connector for processor 2 (if installed) to verify that the problem is not the processor or the memory module connector.

  8. (Trained technician only) Replace the system board.

Multiple memory modules in a channel identified as failing​

Note
Each time you install or remove a memory module, you must disconnect the server from the power source; then, wait 10 seconds before restarting the server.
  1. Reseat the memory modules; then, restart the server.
  2. Remove the highest-numbered memory module of those that are identified and replace it with an identical known good memory module; then, restart the server. Repeat as necessary. If the failures continue after all identified memory modules are replaced, go to step 4.
  3. Return the removed memory modules, one at a time, to their original connectors, restarting the server after each memory module, until a memory module fails. Replace each failing memory module with an identical known good memory module, restarting the server after each memory module replacement. Repeat step 3 until you have tested all removed memory modules.
  4. Replace the highest-numbered memory module of those identified; then, restart the server. Repeat as necessary.
  5. Reverse the memory modules between the channels (of the same processor), and then restart the server. If the problem is related to a memory module, replace the failing memory module.
  6. (Trained technician only) Install the failing memory module into a memory module connector for processor 2 (if installed) to verify that the problem is not the processor or the memory module connector.
  7. (Trained technician only) Replace the system board.

Attempt to change to another DCPMM mode fails​

After the DCPMM mode is changed and the system is successfully restarted, if the DCPMM mode stays the same instead of being changed, check the DRAM DIMMs and DCPMM capacity to see if it meets the requirement of the new mode (see DC Persistent Memory Module (DCPMM) setup).

Extra namespace appears in an interleaved region​

If there are two created namespaces in one interleaved region, VMware ESXi ignores the created namespaces and creates an extra new namespace during system booting. Delete the created namespaces in either the Setup Utility or the operating system before the first booting with ESXi.