Cluster status

News and updates

  • OpenCode on Chimera (2026-10-06): OpenCode, an open-source AI coding agent, is now available on Chimera. It works with the models of the university's LiteLLM gateway and comes connected to the Chimera Slurm and FileCompress MCP tools, with a ready-made hpc-operator agent for cluster work.

    • Install it once by running /work/_software/opencode/install-launcher in a JupyterLab terminal, then restart your JupyterHub server.
    • Use it from any terminal with opencode, or open it from the new opencode tile in the JupyterLab launcher.
    • A LiteLLM API key is optional at installation; the installer shows where to request one, and you can add it later.

  • New Claude Science version 0.1.53 (2026-09-29): A new version of Claude Science is available. To switch to it, start a fresh JupyterHub session with at least 16 GiB of memory, run /work/_software/claudescience/install-launcher in its terminal before opening a Claude Science tile, and then restart your server (see Update).

    • Until you run the installer, your tiles keep the version you have now.
    • Before the new version first runs on your data, the installer backs up both tiles to /work/<username>/.claude-science-backups/. It takes seconds and almost no extra space.
    • Each tile writes its own log, ~/claude-science_<jobid>.log.

  • Cluster is back online (2026-09-17):

    • If you find any issue, please let us know.
    • Gpulab partitions did not change, but the system underneath is now same as in the rest of chimera cluster.
    • Front end and control node are now separated and (hopefully) in HA.
    • Now running on Rocky Linux 9.8 and SLURM 26.05.4.
    • Old GPU lab boxes volta0[1-3,5] are online with updated old gpulab system.
    • Known issues:
      • OpenMPI 5.0.8 on infiniband causes SIGFPE when enabled

  • New communication channel (2026-09-17): All cluster users are now contacted from the new email address support@hpc.cuni.cz, which is the default communication channel from us to you as well as from you to us, see Contacts.

Monitoring The Cluster State

  • Web browser: Real-time status of the cluster can be monitored here.
  • freenodes: A command-line tool on Chimera providing node and partition-oriented load statistics. It was developed for the use on cluster Snehurka - mathematical section of MFF UK. For details visit the help page.
  • smypartition: Displays your available accounts and partitions.
  • sview: A graphical tool for real-time load monitoring; requires SSH X11 forwarding (ssh -X). Sometimes this tool can consume a lot of memory on the headnode so please don't use it unless you know what you're doing.
  • squeue --me To see the status of only your jobs running on the cluster.