跳至主要内容

Water leak problems (Direct Water Cooling Module)

Use this information to resolve water leak problems with the Direct Water Cooling Module (DWCM).

If the water is observed outside of the chassis, make sure the power supplies have been disconnected. If no water is observed outside of the chassis, but there is a suspicion of a water leak in the chassis, complete the following steps to determine the source of the leak. The GPU and CPU water loops are equipped with leakage sensors to assist in detecting any water leaks.
Note
A small leak may not be detected by either leakage sensor, so visual confirmation might be necessary.

Suspicious leakage symptoms​

The following situations might occur due to leakage problems:

  • Processor over temperature error indicated by the System Error "!" LED being solid ON at the front of the server

  • The green LED on the leakage sensor will remain solid when there’s no leakage and will blink at 1 Hz if a leak is detected. Note that the GPU leakage sensor LEDs are only visible when the system is connected to AC power, while the CPU leakage sensor LED can be seen without fully disassembling the server

  • The server shut down unexpectedly

    • If a leak is detected, the system will DC power off and block power permissions. It will not power on again until the leakage issue is resolved.

    • A damaged or pinched cable on the “leak rope” side of the sensor box may lead to system shutdown.

  • If there is an installation error with the leakage sensor, the system will continue to operate, as this error is unrelated to any actual leakage. The two primary causes of this issue are:

    • A damaged cable on the ‘Power’ side of the leak sensor box, located between the sensor box and the board connector.

    • The cable connector is unplugged.

  • A Lenovo XClarity Controller event may report the following events:

    • FQXSPUN0019M: Sensor [SensorElementName] has transitioned to critical from a less severe state. This event indicates there could be leakage detected.

    • FQXSPUN0038J: Sensor [SensorElementName] has indicated a install error.

  • If either a "leak detect" or "sensor install error" persists, the affected water loop will need to be replaced

Complete the following steps in order until you are able to isolate the cause of the potential leak:

  1. Check Lenovo XClarity Controller messages to see if any leakage warnings have been reported. See XClarity Controller events for more information.

  2. Remove the server from the rack, and place it on a stable work surface. See Remove the server from rack.

  3. Locate the water loop indicated by the message.

  4. Use a flashlight to visually inspect the system board and CPU complex for any moisture.

  5. Check the water loop for any moisture.

  6. If you identify the problem in the steps above, replace the water loop (see Remove the Lenovo Neptune(TM) Processor Direct Water Cooling Module).
    Note
    It is important to visually inspect the bottom of the chassis with a flashlight prior to installing the DWCM into the chassis.

  7. After replacing the water loop, run AC cycle and check if the event has been deasserted.

  8. If you are unable to identify the problem, run AC cycle and check if the issue still exist. Contact Product Engineer for the further assistance.