sview

Start a GUI to view and modify the state of Slurm

$ sview

sview [OPTIONS]

-a, --all
    Display all objects, including hidden or inactive ones.

-b, --brief
    Display brief information for relevant objects.

-c, --clusters
    Display information about configured clusters.

-C <file>, --css <file>
    Use a CSS file for output formatting (primarily for web-based views if applicable).

-d, --debug
    Enable debug logging for more verbose output.

-f <format>, --format <format>
    Specify the output format using a comma-separated list of field names.

-h, --help
    Display a help message and exit.

-i <seconds>, --interval <seconds>
    Set the refresh interval in seconds for the interactive display.

-j, --jobs
    Display job information (often the default view).

-n, --nodes
    Display node information within the cluster.

-N, --no-colors
    Disable colored output for terminals that do not support it or prefer monochrome.

-p, --partitions
    Display partition information.

-q, --queue
    Display queue information (jobs awaiting execution).

-s, --state
    Display an overall cluster state summary.

-u, --users
    Display information about users and their jobs.

-v, --version
    Display the sview version number and exit.

-w <width>, --width <width>
    Set the display width, overriding auto-detection.

-x, --no-pager
    Do not use a pager for output, sending all output directly to stdout.

-y <key>, --sort <key>
    Sort the display by the specified field key.

-z <tz>, --timezone <tz>
    Set the display timezone for time-related fields.

sview is a command-line utility for real-time monitoring and introspection of a SLURM workload manager cluster. Similar in concept to the top command, sview provides a dynamic, interactive display of various cluster components.

It allows users and administrators to quickly assess the status of jobs, nodes, partitions, and user activity within the SLURM environment. The display updates periodically, showing current states, resource utilization, and other critical metrics. Users can switch between different views (e.g., jobs, nodes, partitions, users) and apply sorting and filtering directly within the interactive interface, making it an invaluable tool for cluster management and troubleshooting. It provides a quick overview without needing to run multiple static sinfo or squeue commands.

sview requires a functional SLURM Workload Manager installation to operate. The information displayed is only as accurate as the slurmctld daemon provides. Users need appropriate permissions to access certain cluster details. Its primary utility is within SLURM environments; it has no function outside of them.

While running sview, users can interact with the display using various keyboard shortcuts. Pressing 'h' or '?' typically brings up an in-application help screen detailing available commands, such as 'q' to quit, 'j' for job view, 'n' for node view, 'p' for partition view, and commands for sorting and filtering output.

sview offers distinct display modes accessible via command-line options or interactive commands. The default mode often shows jobs. Other common modes include viewing cluster nodes, active partitions, a summary of the cluster state, or details about users. Each mode presents relevant metrics and statuses tailored to that specific aspect of the SLURM cluster.

sview is an integral part of the SLURM (Simple Linux Utility for Resource Management) Workload Manager, which originated at Lawrence Livermore National Laboratory (LLNL). First released in 2002, SLURM was developed to manage and schedule large-scale computing clusters. sview evolved as a real-time, interactive monitoring tool within the SLURM suite, providing a dynamic alternative to static query commands for administrators and users to observe the fluctuating state of the cluster, jobs, nodes, and partitions. Its design reflects the need for immediate insight into cluster operations for efficient resource management and troubleshooting.

sinfo(1), squeue(1), scontrol(1), sacct(1), slurm(1), top(1)