Multi L4 GPU server freeze

When I run multiple DeepStream Docker instances related to machine vision with 2 Nvidia L4 GPUs, I frequently encounter system crashes on Ubuntu, and the likelihood increases under high server load. During these incidents, the GPU temperature remains around 87 degrees, which appears to be within the normal operational range.
Additionally, the same issue has occurred on another server equipped with 8 L4 GPUs. It’s worth mentioning that this Docker instance has not yet encountered the same freezing issue when running on T4 and other NVIDIA graphics card servers. How should I resolve this problem?

Description: Ubuntu 22.04.3 LTS
Release: 22.04
Codename: jammy

nvidia-bug-report.log.gz (1.2 MB)
syslog-20240315.log (1.5 MB)