Recently, I had some cooling issues with Mellanox Infiniband NICs. I installed passive cooling NICs to water-cooling servers. Due to lack of airflow, those NICs start to throttle after reaching 80°C. I had to check its temperature frequently so that bandwidth degradation does not interfere with my experimental results. Mellanox provides a tool mget_temp that… Continue reading How to check temperature of Mellanox Infiniband NIC with non-root users (Harder than you think!)