SINFO(1)                        Slurm Commands                        SINFO(1)

NAME
       sinfo - View information about Slurm nodes and partitions.

SYNOPSIS
       sinfo [OPTIONS...]

DESCRIPTION
       sinfo reports the state of partitions and nodes managed by Slurm.
       By default each output line describes a group of nodes in one
       partition that share the same state; nodes are shown as a
       compressed hostlist such as node[01-04].  The default partition is
       marked with an asterisk.

OPTIONS
       -h, --noheader
              Do not print the header line.

       -l, --long
              Print more detail, below a timestamp line.

       -N, --Node
              Print one line per node (per partition the node is in).

       -n, --nodes=<nodes>
              Report only on these nodes; hostlist expressions are
              accepted.

       -o, --format=<output_format>
              Choose the columns.  Each field is written %[.][size]type; a
              "." right-justifies the value and size sets the width.
              Useful types include:

              %P  partition (with * on the default)   %R  partition name
              %a  availability (up/down)              %l  time limit
              %D  number of nodes                     %N  node list
              %t  state, short form                   %T  state, long form
              %c  CPUs per node                       %C  CPUs A/I/O/T
              %m  memory per node (MB)                %G  generic resources
              %E  reason                              %u  user of reason
              %H  reason timestamp                    %f  features
              %F  nodes A/I/O/T                       %z  sockets:cores:threads

              The default is "%9P %.5a %.10l %.6D %.6t %N".

       -p, --partition=<partition>
              Report only on these partitions.

       -R, --list-reasons
              List nodes that are down, drained or not responding together
              with the reason, the user who set it and when.

       -s, --summarize
              One line per partition with node counts in the form
              allocated/idle/other/total.

       -t, --states=<states>
              Report only nodes in these states (comma list), for example
              idle, alloc, mix, drain, down.

       -V, --version
              Print the Slurm version.

NODE STATE CODES
       Short forms as printed by %t (long forms by %T); a trailing *
       means the node is not responding to the controller.

       alloc    all CPUs are in use by jobs              (allocated)
       mix      some CPUs are in use                     (mixed)
       idle     no jobs are running                      (idle)
       drng     draining: no new jobs, current ones run  (draining)
       drain    drained: no jobs, none will start        (drained)
       down     unavailable for use                      (down)
       inval    registered with an invalid configuration (inval)

EXAMPLES
       Show every node on its own line:

              $ sinfo -N -l

       List nodes that are out of service and why:

              $ sinfo -R

       Show partition, node list, state and GPUs per node without a
       header:

              $ sinfo -h -o "%P %N %t %G"

       Show only idle nodes in one partition:

              $ sinfo -p batch -t idle

SEE ALSO
       squeue(1), scontrol(1), sacct(1)

Slurm                             GreenRay AIO                        SINFO(1)
