Pod Status of the NPU-Exporter Is CrashLoopBackOff

Symptom

The pod status of a component is CrashLoopBackOff. The following information is displayed in the log:

[ERROR]    2022/10/25 20:33:10.743372 1       x509@v0.0.5/certificate.go:193    psFile and psFileBk file is not exist
[ERROR]    2022/10/25 20:33:10.743444 1       npu-exporter/main.go:109    file does not exist

Causes

By default, the NPU-Exporter is started in HTTPS mode. Before starting the NPU-Exporter, you need to import the HTTPS certificate. The error cause is that the HTTPS certificate is not imported.

Solution

Use either of the following methods:

• If the component is deployed in container mode, modify the NPU-Exporter startup YAML file, add the startup parameter --enableHTTP (setting it to true), and deploy the NPU-Exporter again. The component will be started in HTTP mode.
• Import the certificate and deploy the NPU-Exporter again. For details, see Importing a Certificate and KubeConfig File.