We are not saving a pending job's task distribution, so after restarting
slurmctld select/cons_res was over-allocating resources based upon an uninitialized distribution value. Since we can't save the value without changing the state save file format, we'll just set it to the default value for now. This will result in an incorrect task distribution for jobs that had a task distribution that was not the default and were pending when the slurmctld daemon restarted, but at least resources will not be over-allocated.
Loading
Please register or sign in to comment