Jobs Management

  2. Home
  4. Docs
  6. Jobs Management
  8. Basic Commands

Basic Commands

Two of the most basic commands of slurm scheduling system are sinfo and squeue. For extended documentation, someone could refer to the corresponding webpages (sinfo, squeue). A brief summary of these two commands follows. For jobs monitoring purposes through advanced use of sinfo and squeue, users can rely on the article Jobs Monitoring of this documentation.

sinfo:view of current state of nodes and partitions

The following picture depicts the output of sinfo command:

  • Column PARTITION: cluster partitions are shown. The asterisk in “batch” partition means that batch is the default partition (if a partition is not defined in a user’s job, then the job is allocated to the batch partition).
  • Column AVAIL: the availability of the partition is provided (expected values  up/down).
  • Column TIMELIMIT: the time duration that a job is allowed to run. In fast partition, this time is limited to 12 hours.
  • Column NODES: the number of nodes at a specific state. For example, four nodes (node 4,5,6 and 10) are at “alloc” state.
  • Column STATE: the current operation mode.
 An idle partition/node is available for allocation and job execution
 On an allocated partition/node a job or jobs are already running so allocation is impossible
A down partition/node is unavailable due to technical reasons
On a mix partition/node a job or jobs are already running but there are  remaining resources (cores/memory) so that new jobs can be allocated.
  • Column NODELIST: the set (names) of nodes at a specific state.

squeue: view of current state about the slurm scheduling queue

In the following pictures the output of squeue is illustrated.

  • Column JOBID: the id of the job. Jobid is automatically given by slurm controller.
  • Column PARTITION: the partition where the job is running.
  • Column NAME: the name of the job (assigned by user).
  • Column USER: the name of the user whose job is running.
  • Column ST: the state of the job. Typical values are: R(running),  PD (pending), CG (completing), CD (completed)
  • Column TIME: the elapsed time since the beginning of execution
  • Column NODELIST(REASON): The node(s) where the job is running or the reason that the job is pending for execution for. Typical values of REASON:  Resources (the job is waiting for resources to become available), Priority (One or more higher priority jobs exist for this partition), Dependency (This job is waiting for a dependent job to complete)