Why do qos_gpu-t4 jobs have a PENDING status for the reason of QOSGrpCpuLimit or QOSGrpGRES?

As indicated in our documentation concerning the GPU QoS , the totality of qos_gpu-t4 jobs executed simultaneously by all the users cannot occupy more than 32 converged nodes.

When this limit is reached, your jobs will be PENDING for the reason of QOSGrpCpuLimit or QOSGrpGRES.

Important: Do not delete them. They will be executed as soon as possible.