Supports multiple clusters in one web interface. Scales to thousands of nodes and millions of jobs. Supports heterogeneous clusters with and without node sharing.
Comes with a ready to use Slurm integration. Can be integrated with any batch job scheduler via a REST API.
Responsive web interface with job specific, node, user, and system views. Provides metric plots, aggregate statistics, roofline plots and ressource table, and access to job script. Fully user configurable.
Access to configurable set of metrics including hardware performance counter data. Comes with powerful HPC centric node agent, but can also be integrated with other node agent solutions.
Supports local accounts, LDAP, and KeyCloak OpenID Connect. Can be integrated with existing user portals using JWT based authentication.
Supports roles for users, project managers, support personnel, and administrators. Users can only see their own jobs.
Powerful sorting of jobs according to all job metadata attributes. Filter for job metadata and aggregate metric data attributes.
A unified search bar allows to search for job ids, job names, project ids, usernames, and names.
Jobs can be tagged. Job tags are grouped by type and can have a configurable visibility using a scope attribute.
TU Darmstadt, Ruhr University Bochum, University Duisburg-Essen