Skip to content

v5.0.0

Choose a tag to compare

View all tags
@rezib rezib released this 27 May 07:28
· 313 commits to main since this release
v5.0.0
This tag was signed with the committer’s verified signature. The key has expired.
rezib Rémi Palancher
GPG key ID: 5B8B293B68C6FEA9
Expired
Verified
Learn about vigilant mode.
d6f4662

Important

Please read update guide for updating from Slurm-web 4 to Slurm-web 5!

The default authentication mode to slurmrestd has been changed from local to jwt because of some changes in Slurm 25.05. This is a breaking changes for some setups. Some manual steps must be performed for this update, instructions are provided in the link above.

Added

  • Support Slurm 25.05 and Slurm REST API v0.0.41 (#541).
  • GPU resources utilization monitoring (#258).
  • frontend:
    • Dark mode support (#278).
    • Display cluster total memory size and number of GPUs in dashboard stats.
    • Display number of GPUs per node in resources page and node details page.
    • Display number of allocated GPU and allocation percentage in node details page.
    • Display number of GPU requested or allocated in jobs list.
    • Display number of GPU requested and allocated in job details page.
    • Add buttons to display GPU status metrics in dashboard.
    • Reproduce Slurm job status logic for jobs badges (#404).
    • Reproduce Slurm node status logic for node status and allocation badges with icons to represent status flags (#405).
    • Add node filter for error and fail states.
    • Display version in below logo in main menu (#568).
    • Display suspended, preempted, deadline, node fail, boot fail, out of memory jobs in dashboard chart.
    • Display fail and error nodes in dashboar chart.
  • agent:
    • Add total quantity of memory and number of gpus in cluster stats response.
    • Add metrics for GPU by state and total number of GPU.
    • Support querying GPU metrics.
    • Add missing jobs base states in collected metrics: SUSPENDED, PREEMPTED, NODE_FAIL, BOOT_FAIL, DEADLINE and OUT_OF_MEMORY.
    • Add ERROR and FAIL node states in collected metrics.
  • gateway: Add version in configuration file generated for frontend component.
  • conf:
    • Select gres and gres_used attributes by default in slurmrestd nodes list and indiviual node responses.
    • Select gres_detail, sockets_per_node, tasks, tres_per_job, tres_per_node, tres_per_socket, tres_per_task attributes by default in slurmrestd jobs list and Slurm controller invidual job responses.
  • docs:
    • Mention support of Fedora 42.
    • Mention GPU support feature in overview page.
    • Mention GPU metrics in metrics configuration reference documentation.
    • Mention dark mode support feature in overview page.
    • Mention new --with-slurm option in slurm-web-gen-jwt-key manpage.
    • Add update guide page with special notes for manual steps required to update from v4 to v5.

Changed

  • agent:
    • Run agent as slurm-web system user by default.
    • Emit warning log entry at launch when slurmrestd local authentication method is used.
  • frontend:
    • Migrate to Tailwind CSS v4 (#449). Note this breaks support of old browsers versions, it requires Safari 16.4+, Chrome 111+, and Firefox 128+.
    • Convert nodes memory size in GB or TB in resources page and node details page.
    • Appearance of jobs sort button to visually separate its double role, change sort order and select sort criterion.
    • Bump heroicons dependency to version 2.2.0.
    • Group drain and draining node state filters.
  • genjwt: Give ownership of Slurm-web JWT signing key to slurm-web system user only by default. Read ACL is added for slurm system user only if --with-slurm option is provided explicitely.
  • connect: Emit warning log entry when slurmrestd local authentication method is used.
  • ldap-check: Return exit code 1 on LDAP authentication error.
  • conf:
    • Bump [slurmrestd] > version default value from 0.0.40 to 0.0.41 in agent configuration for compatibility with Slurm 25.05.
    • Change default value of [slurmrestd] > auth agent parameter from local to jwt.
  • docs:
    • Replace mention of Slurm REST API version v0.0.40 by v0.0.41.
    • Mention requirement of Slurm >= 24.05 and dropped support of Slurm 23.11.
    • Update quickstart guide to setup slurmrestd JWT authentication, automatic token management mode on Unix socket by default.
    • Update installation guide and quickstart guide to mention new separate slurmweb-5 repository.
    • Update troubleshooting guide with different curl command for slurmrestd JWT authentication and TCP/IP socket.
    • Mention slurmrestd local authentication is deprecated in architecture and slurmrestd configuration pages.
    • Update slurmrestd configuration guides to reflect change to jwt authentication method by default.
    • Update configuration reference documentation.

Fixed

  • agent: Fix AttributeError with prometheus_client.registry.Collector on el8 (#548).
  • frontend:
    • Rendering of nodes grid in resources page in presence of long node names.
    • Update dependencies to fix CVE-2025-46565 (vite).
  • ldap-check: Report and return error instead of crashing when LDAP URI is not defined in configuration.

Removed

  • Drop support of Slurm 23.11 and Slurm REST API v0.0.40.
  • docs: Drop support of Fedora 40.
Assets 2
Loading
0 Join discussion