## December 15th, 2025
* Agenda
* Status
* 25.11 is now the default branch on Charmhub
* `set-node-state` action is now implemented in `slurmctld` charm.
* Working on `default-state` and `default-reason` configuration options.
* Check out the new Glossary on the documentation!
* https://documentation.ubuntu.com/charmed-hpc/latest/reference/glossary/
* Adding docs in January for `ldap-integrator`, replacing the current glauth-k8s section.
* Open floor
* Charmed HPC LinkedIn
* Nvidia acquires SchedMD
* Contributing to Debian HPC packages
## December 10th, 2025
* Agenda
* Status updates
* Created `libnss-slurm` package
* Not fully integrated into the Slurm charms as we need to create a parser for nsswitch.conf.
* Any ideas, let us know!
* Documentation updates
* New "integrate" section
* Charmhub updates
* CIUK
* Open floor
* CUDA on Resolute
* Slurm factory progress
## November 26th, 2025
* Agenda
* Documentation changes
* Updating extensions
* CIUK is near: 3-5 Dec
* Any demo videos?
* LDAP integrator progress
## November 19th, 2025
* Agenda
* Updating Charmhub
* LDAP proxy is almost complete
* `gres.conf` bug is fixed
* Now developing on Slurm 25.11
* Action items
* Deep dive on `slurm-bridge`
* Kyle propose a LinkedIn page for Charmed HPC
## November 12th, 2025
* Agenda
* Sprint updates - We're back!
* Good path forward for Charmhub cleanup
* Need to delete the default track
* This should get rid of all metadata
* Setup Slurm tag for 23.11
* Identity discussion
* We're creating an LDAP proxy charm
* Will work nicely with existing SSSD charm
* GLAuth won't be part of our identity stack
* Issues with Slurm 25.05
* Slurmdbd and Slurmctld issue
* Both services generate different cluster IDs
* This causes a mismatch which breaks deployment
* Slurm 25.11 fixes the issue
## October 29th, 2025
* Agenda
* Status updates
* pyfakefs bug fix affecting slurmctld HA unit tests
* [Release policy spec](https://github.com/charmed-hpc/specs/blob/main/specs/UHPC%20003%20-%20Release%20policy%20and%20notes%20for%20Charmed%20Slurm/uhpc-003.md) merged
* No community call next week - Canonical team will be sprinting
* Open floor
* Update Latest on Charmhub - e.g. remove CentOS logo and older Ubuntu revisions
## October 22nd, 2025
* Agenda
* Status updates
* CIUK talk
* UbuCon LA
* Internal testing on MAAS
* Migration to documentation.ubuntu.com
* New URL: documentation.ubuntu.com/charmed-hpc/latest/getting-started
* New tracks on Charmhub
* New issues opened
* https://github.com/charmed-hpc/slurm-charms/issues/151 (needs triage)
* https://github.com/charmed-hpc/slurm-charms/issues/149 (PR open to improve things)
* Open floor
* Login node vs sackd terminology (open discussion on org)
* Node exporter discussion (needs more investigation)
## October 8th, 2025
* Agenda
* Status updates
* 26.04 cycle planning
* Removing `glauth` from the documentation
* Release infrastructure for Charmhub
* Open floor
* ...
## October 1st, 2025
* Agenda
* Status updates
* Tutorial is online!
* https://canonical-charmed-hpc.readthedocs-hosted.com/latest/getting-started/
* LDAP/Identity updates
*
* Open floor
* ...
## September 24th, 2025
* Agenda
* Status updates
* Cycle planning
* Identity stack
* Tutorial preview
* Charmhub preparation
* Open floor
## September 17th, 2025
* Agenda
* Status updates
* Conferences!
* DevConf.US
* CIUK (submitted abstract)
* Open floor
* Slinky
## September 10th, 2025
* Agenda
* Status updates
* HA support has been merged!!!! 🥳🥳🥳
* A lot of complexity abstracted away, handled by Juju and the charms.
* Tutorial [draft](https://github.com/charmed-hpc/docs/pull/80)
* Polishing the cloud init script.
* Apptainer documentation
* Setup and how to use.
* Preparing talk for CIUK
* Open floor
* ...
## September 3rd, 2025
* Agenda
* Status updates
* Tutorial [draft](https://github.com/charmed-hpc/docs/pull/80)
* Experimental branch updates
* Merged into main
* HA coming soon!!
* Charmhub track for the Slurm charms
* Open floor
## August 27th, 2025
* Agenda
* Status updates
* Tutorial
* Experimental branch updates
* HA coming soon!!
* OCI workload scheduling demo
* Open floor
* ...
## August 20th, 2025
* Agenda
* Slurm interfaces refactor
* Working!
* Open floor
* ...
## August 13th, 2025
* Agenda
* Documentation updates
* HA docs in progress
* Contributing to Charmed HPC
* Open floor
* ...
## August 6th, 2025
* Agenda
* Slurm backports
* Open floor
* debian patches
## July 30th, 2025
* Agenda
* GCP work
* Open floor
* ...
## July 23rd, 2025
* Agenda
* Dynamic nodes demo
* Experimental branch - sandbox for HA and other work
* GCP work
* Open floor
* ...
## July 16th, 2025
* Agenda
* Slurm HA
* Dynamic nodes
* Open floor
* Release policy
* Two tentative release policies.
* This cycle: Charmed Slurm
* Next cycle: Charmed HPC
* What’s the difference?
* Charmed Slurm is infrastructure. You get the workload scheduler with all the fancy integrations for observability and scale up and down, but you’re responsible for providing application stack and managed identities.
* Charmed HPC is “managed applications” + Charmed Slurm. Charmed HPC is the full application suite with managed identity and workload support.
* Skeleton of published release policy notes:
* Outline
* What is contained within this page, summary of content, where to go announcements tracking.
* Foresight: Link to upgrade instructions once we actually support upgrades.
* Recent releases with support lifecycle.
* Upcoming release roadmap (no target date cause that’s pressure)
* Release and support policy
* Major release (new version of Slurm, breaking changes, new base OS)
* Minor release (new features, but no breaking changes)
* Patch release (bug and security fixes)
* Include note - “plz `juju refresh` if you want the latest security and bug fix updates. Otherwise die”
* Foresight: Feature deprecation
* What’s new in Charmed Slurm __?
* High-level overview of new features/major changes.
* Notable bug fixes.
* Link to more in-depth release notes page
* Security Policy
* Link to the general Charmed Slurm security policy.
* Skeleton of “in-depth” release notes page.
* Summary of what’s new in the release.
* Summary of requirements to deploy Charmed Slurm
* New features + improvements (if any)
* Known issues
* Patched CVEs
* Bug fixes (link to issues on GitHub)
* Upgrade instructions (if any)
* How do we automate this so we do not become professional Markdown programmers?
* Tie issues to release milestones
* Already have the issues tags + types for categorization
* Make sure CVEs are accurately tracked within the project.
* Subscribe to Slurm security mailing list if there is one.
* New features - document via specs.
* “Release policy”
* Something internal to the team, not necessarily something we need to make publicly available. Can live within a spec, but doesn’t need to go on RTD.
Skeleton
* What’s included in a stable release?
* List supported artifacts with link to source code + issue tracker
* Table outlining release structure.
* E.g. stable release of the Slurm charms are published to 25.04/stable on Charmhub
* ^ Mirror this bit in the release notes. E.g. how do you actually pull Charmed Slurm 25.04
* PPA to get supported Slurm packages.
* Any Terraform plans for reference deployment (maybe, homies could just pull from the correct channel)
* How long do we commit to supporting a release?
* 18 months
* When do we release?
* A month after the newest version of Slurm is published?
* Infrastructure for supporting releases.
* Already have the publishing + test pipeline
* Create Git branches for every major version of Slurm.
Merge PR against the relevant feature branch and bubble up from there.
Nice to have: GitHub bot that enables us to promote charms directly from call-for-testing issues.
## July 9th, 2025
* Agenda
* Status updates
* New `specs` repository
* {%preview https://github.com/charmed-hpc/specs %}
* Draft release policy spec:
{%preview https://github.com/charmed-hpc/specs/pull/1 %}
* GitHub Discussions clean-up
* Filesystem client + HA work
* Dynamic nodes work
* Open floor
* ...
## July 2nd, 2025
* Agenda
* Status updates
* GCP Terraform module for Charmed HPC
* HA integration tests
* Filesystem client work
* Vendoring `slurm-ops` into Charmed HPC + Dynamic nodes
* Open floor
* Node Exporter Operator
* `db-uri` PR. Please review.
## June 25th, 2025
* Agenda
* Status updates
* Slurmctld High Availability (https://github.com/charmed-hpc/slurm-charms/pull/120)
* {%preview https://hackmd.io/e08R0gDPR8eU4OMaSxsGCA %}
* Internal packages (https://github.com/charmed-hpc/slurm-charms/pull/121)
* Dynamic node (WIP)
* GCP Terraform (WIP)
* Open floor
* High Availability failover demo
## June 18th, 2025
* Agenda
* Status updates
* `commitlint` demo
* Org-README and [CONTRIBUTING guide](https://github.com/charmed-hpc/.github/pull/19) updates
* Dynamic compute node spec: {%preview https://hackmd.io/@ubuntu-hpc/SJU0Xk0Qeg %}
* Currently working on this specification to introduce dynamic compute nodes in Charmed HPC. This will replace the current way we enlist compute nodes in Charmed HPC, so please offer your input.
* Forseen benefits include:
> _See spec for full description of benefits_ :eyes:
* No more _service.py_ wrapper for the slurmd daemon.
* Compute node configuration - e.g. `NodeName=... Gres=...` - will no longer be tracked in integration data.
* Node state will be controlled directly using `scontrol` rather than through integration data updates. E.g. `node-configured` will call `scontrol update nodename=<node name> state=up` rather than wait for `RelationChanged` events to be handled.
* __This will greatly simplify how we manage Slurm's configuration in the charms, and it will make it easier for us to add new features to the slurmd/slurmctld charms.__
* Open floor
* Public specs for new features and major changes.
## June 11th, 2025
* Agenda
* Status updates
* New PR labels
* Documentation
* Org-level README: https://hackmd.io/@ubuntu-hpc/SJAM3pHmgx
* Contributing guide: https://github.com/charmed-hpc/.github/pull/19
* Ongoing Apptainer work
* AWS modules
* Open floor
## June 4th, 2025
* Agenda
* Status updates
* Apptainer
* AWS Terraform work
* Documentation
* Open floor
## May 28th, 2025
* Agenda
* Status updates
* Filesystem support for GCP, AWS, OCI
* OCI support
* Community engagement - linting tools and automation
* Open floor
## May 21st, 2025
* Agenda
* Status updates
* `slurmutils` v1 API is out now.
* Strict typing for Slurm configuration
* More reliable callbacks
* Better API for writing new editors
* Documentation updates
* Will be adding more usage howtos
* Open floor
## May 7th, 2025
* Agenda
* Status updates
* HA support work
* Slurmutils changes
* Documentation updates
* GLAuth work
* Terraform changes
* Open floor
* No community call next week.
* Canonical team are attending the company Engineering Sprint in Frankfurt.
## April 23th, 2025
* Agenda
* Status updates
* HA support work
* Slurmutils changes
* Documentation updates
* GLAuth work
* Open floor
## April 16th, 2025
* Agenda
* Status updates
* HA support work
* Slurm charm interface work
* New docs!
* GLAuth work
* Open floor
## April 9th, 2025
* Agenda
* Status updates
* Slurm HA support
* Migrating from auth/munge to auth/slurm
* slurmutils work
* Documentation updates
* Open floor
* 
## April 2nd, 2025
* Agenda
* Move meeting time due to DST?
* Moving to 16:30 UTC!
* Status updates
* Slurm HA support
* Migrating from auth/munge to auth/slurm
* slurmutils work
* Open floor
## Match 26th, 2025
* Agenda
* Status updates
* Documentation updates
* Benchmarking tests
* UV + Just = ♥️ (Demo next week!)
* Open floor
* Vantage Compute
* New Slurm environment deployment for client
* Admin infra (slurmctld, slurmdbd, mysql) in AWS
* Compute nodes on-site
* Wants to use RDS instead of mysql instance - database proxy for slurm charms?
* HA for slurmctld and slurmdbd
* nfs options
* AWS integration
* EFS proxy charm?
* See: https://github.com/canonical/anbox-cloud-nfs-operator. Can use the same logic for the `filesystem-charms`
* Discuss Development
* Database proxy
* nfs proxy for EFS
* Check if filesystem-client can just immediately
* HA slurmctld and slurmdbd
* Use NFS + filesystem-client to configure HA slurmctld.
* Jupyterhub + SlurmSpawner
## March 19th, 2025
* Agenda
* Status updates
* Documentation updates
* Discussion:
* Charm actions redesign
* https://github.com/orgs/charmed-hpc/discussions/16
* Please comment if interested.
* Release policy
* Open floor
* New COS dashboards
## March 12th, 2025
* Agenda
* Status updates
* Documentation updates
* `uv`
* SCALE 22x
* Open floor
## March 5th, 2025
* Agenda
* Status updates
* Maintenance work
* Fixed https://github.com/charmed-hpc/slurm-charms/issues/85
* Need to reduce the number of TICS analysis to once per week.
* COS dashboards improvements
* Looking into improving the dashboards offered on Grafana to be a bit more useful.
* Benchmarking
* Created new repository for benchmarks https://github.com/charmed-hpc/charmed-hpc-benchmarks.
* Documentation
* PR template merged!
* Since the PR template lives in the .github repo, it applies to all repos. Need to override this default for the docs repo.
* Cory:
* Working with GSoC 2025 to enable AMD GPU acceleration in various Debian Science packages.
* NVadsV710_v5 is supported on Azure (preview)!
## February 26th, 2025
* Agenda
* Slurm charm latency issues and design aspects
* Looking at platform stabilization
* Design for how Slurm units communicate
* Promising results so far - more control over how data is sent
* Benchmarking/test suite
* Using ReFrame to benchmark Charmed HPC
* Small script written in Bash that deploys and benchmarks IO (NFS perf)
* New PR template - example on .github repo
* Templates that will be used whenever a new PR is opened
* Docs check - user confirms that docs changes are in progress or not needed
* Open floor
* Discussion opened for improving UX for Slurm charms: https://github.com/orgs/charmed-hpc/discussions/16
* Redesign ideas (requires sign in): https://hackmd.io/@ubuntu-hpc/BkM7BZrckl
## February 19th, 2025
* Agenda
* Status updates
* Code review - looking at the following PRs:
* https://github.com/charmed-hpc/hpc-libs/pulls
* https://github.com/charmed-hpc/slurm-charms/pulls
* Charmed HPC testing
* MPI, fio, GPU burn
* `hpctestlib` - https://github.com/reframe-hpc/reframe/tree/develop/hpctestlib
* Slurm charm testing
* Demo time!
* Open floor
## February 12th, 2025
* Agenda
* Status updates:
* Maintenance work
* Added SECURITY.md
* Filesystem client + NFS
* https://github.com/charmed-hpc/charmed-hpc-terraform/pull/11
* Documentation
* How-to guides getting some love!
* Added a how-to on initializing the cloud environment.
* K8S + machine cloud setup.
* Classified how-tos on sections.
* More explanations!
* More reference!
* UX (Dominic)
* Did an extensive investigation of UX for Charmed HPC
* Expected users
* Current shortcomings
* Things to consider
* Should help guide UX decisions in the future.
* Internals
* Working towards having a better set of interfaces for communication between the slurm charms.
* Composable interfaces.
* Secrets for slurm key and munge key.
* Better way to check if relation data is available.
* Demo time!
* GPU + Infiniband + NFS
* It works! :D
## January 29th, 2025
* Agenda
* Status updates:
* Maintenance work (CI)
* Charmcraft 3.3.2 causing integration test failures. Issue filed and raised with the team.
* Identity
* SSSD operator changes merged
* Now integrates with glauth-k8s charm supported by Canonical identity team
* Filesystem client + Lustre
* Azure does not offer packages for Noble for Lustre
* Whamcloud provides 24.04 packages but only for a specific (older) kernel version
* Could package Lustre ourselves with DKMS support but incurs maintainability costs.
* Working on resolution.
* Documentation
* Updates have been made: GPU, filesystem, etc.
* Canonical Open Documentation Academy (CODA) updates coming
* Infiniband
* PR in review. Changes being worked on.
* Discussions:
* Arif on Infiniband subnet management:
* "Is this handled by Azure?" (Yes, Azure handles it behind the scenes)
* Recommendation ~5 years ago was to run OpenSM on metal for larger clusters and not rely on the switches to do subnet management
* Running OpenSM on the controller/slurmctld node may be an option when we get there
* Introductions from Bennett Hunter, University of Chicago HPC cluster manager
## January 22th, 2025
* Agenda
* Status updates:
* Maintenance work
* Identity
* Filesystem client + Lustre
* Documentation
* Infiniband
* Discussions:
* Infiniband detection
* Dominic: Should we install rdma-core by default on slurmd? Would allow us to autodetect Infiniband without much problems.
* Jason: It would be good to first check what can we do using kernel APIs
* Dominic: I'm worried that we won't be as exhaustive as the people maintaining a specialized library to do that.
* Jason: Yeah, maybe that's true. It would be bad to have a double negative where the autodetection code didn't detect IB but the node does have it.
* DOCA OFED vs default kernel drivers
* James: What is the difference between the default drivers and DOCA OFED? Wasn't there a discussion related to that?
* Jason: Yeah, we had the same question. It seems like DOCA OFED has some additional patches and optimizations that aren't included by the default drivers. However, the default drivers are signed and easy to install, so it frees us from the troubles of having to e.g. pass additional parameters to Juju.
## January 15th, 2025
* Agenda
* Status updates:
* Filesystem work
* GPU work
* Identity
* Documentation
* Discussions:
* Should `app_name` be a required argument in our charm Terraform modules, or should we just supply the charm name as the default application name? E.g. application name for `slurmctld` charm is "slurmctld" if an app_name is not supplied to Terraform.
* Set charm name as default to have feature parity with Juju CLI.
## January 8th, 2025
* Agenda
* Happy new year! :tada:
* Status updates:
* Filesystem work
* Bug fixes for inconsistencies in Python 3.9 -> 3.12.
* Working on non-Juju-handled file system support.
* Expecting merged within the next day or so.
* Lustre up next!
* Identity
* Bug with sharing CA certificates: https://github.com/canonical/glauth-k8s-operator/issues/84
* GPUs
* PR open: https://github.com/charmed-hpc/slurm-charms/pull/57
* Before merge, looking to move some functionality into `slurmutils`, add dependencies into the charm (rather than apt installing them), and do some refactoring.
* Discussions
* Arif: from a support perspective, allowing custom drivers (not from archive) can create complications. Possible approaches: allow configuration of a known source for drivers, e.g. a known PPA or other deb repo.