Skip to content

FY 2026 Project Planning

Triaging and prioritizing RCS projects for FY26.

Please add comments to this issue rather than editing directly.

MUST >> WANT >> NICE

  • TOP priority
    • DOCS: Update facilities document https://github.com/uabrc/uabrc.github.io/issues/780 (and more, talk to William)
    • CHEAHA: Migrate Head Node from GPFS4 to GPFS5 cheaha#62 (Slurm Shutdown reqd https://github.com/uabrc/devops-docs/issues/70)
    • CHEAHA: Update Cheaha OS cheaha#55
      • CHEAHA: Cheaha software builds via CICD (next tier priority) #679
    • STORAGE: GPFS5/Ceph Core shared quotas #632
      • CHEAHA: intel-dcb nodes made available on GPFS5 for internal priority (windfall?) use (facilitates calculating shared quotas via policy runs) gpfs-policy#61
    • STORAGE: Scratch policy implement GPFS5 cheaha#34
    • STORAGE: Cutover apps/modules from GPFS4 to GPFS5 gpfs5-migration#208
    • STORAGE: Auto-hydrate all files <= 4 MB into performance tier (resolves conda/git hydration issues)
    • GLOBUS: Migrate Shared Collections from GPFS4 to GPFS5 Mapped Collection #695
    • REDCAP: Redcap deploy from SOM to OpenStack
    • REDCAP: Install Redcap OpenStack extension
    • STORAGE: Fix LTS-Globus bug #608
    • PUBLICATION: Good 2025 conference paper
      • PEARC 2026 paper (Feb 9 due date, unlikely to hit this date) and (May 4 due date for short paper, no changes needed)
      • Supercomputing 2026 paper (April 1 due date, unlikely to be accepted due to topic without tech details)
  • MUST haves (requirements)
    • XNAT: Resolve ongoing data management issues (Update XNAT)
    • GITLAB: Update GitLab OS gitlab#12
      • GITLAB: Update GitLab version #689
        • RCS: Software license renewal and upgrade (patch) planning and reminders/automation #686, Blocked by #689
          • Researcher-Facing software (MATLAB, SAS, STATA, ...)
          • Internal software (OS, Slurm, ...)
    • SYSTEM: Disaggregate LDAP service from Master Node
    • CHEAHA: Disaggregate Slurm service from Master Node
    • STORAGE: Ceph Block/LTS/FS (not GPFS) shared quotas and accounting
    • CHEAHA: Update Slurm
      • CHEAHA: Slurm accounting model and priority queueing to support hardware buy-in
      • CHEAHA: Windfall "ondemand" partition #673 #471
    • CHEAHA: Update Lmod (EasyBuild 5 will require Lmod >=8)
    • OOD
      • CHEAHA: Update OOD (can get to 3.x)
      • CHEAHA: -ntasks-per-socket on OOD apps for correct multi-GPU usage (test OOD 3.x update first)
    • CHEAHA: Node hardware info in slurm.conf (features) rc-data-science/metrics/rc-hardware#2 #230
    • CHEAHA: Deprecate Anaconda3 module in favor of conda-forge (license compliance)
    • CHEAHA: Add Makevars.site file to R modules on Cheaha to manage "Illegal Instruction" errors #708
    • COMMUNICATION: Email sending automation (no more Outlook) #674
    • SECURITY: SSO web service alignment everywhere for MFA and security compliance
    • SECURITY: Get information from XIAS for security
      • (for person X, what is sponsor BlazerID? is X expired? what is X site(s)? is site expired? is URI in site?)
    • SECURITY: Security (POAM) milestones (docs, procedures, etc)
    • SECURITY: HIPAA attestation
    • SECURITY: NIST 800-171 attestation other than NIH (compliant for NIH grants, DUAs go thru OSP, limited to RCS only)
    • SECURITY: Automated account state management (this is very broad) #669
    • SECURITY: Enable and require 2FA for all systems (includes SSH)
    • RCS: Resource entitlement and eligibility tracking (who can have/has which shared allocations) #637
    • BRANDING: Branding all relevant sites (docs, OOD, cloud.rc, etc) #624 #556
    • ADMIN: Internal staff onboarding procedure(s)
  • WANT to haves (higher priority)
    • LTS: Unify individual allocation labels to match blazerid@uab.edu format.
    • CHEAHA: SAS Enterprise Guide OOD app cluster-software#139
    • CHEAHA: Install most recent SAS
    • CHEAHA: Install most recent STATA
    • CHEAHA: Prevent/auto-revert Cheaha TLD permission changes
    • CHEAHA: Refine experience for TMP and LOCAL dirs on Cheaha nodes #678
    • CHEAHA: Cheaha Slurm QoS, priority calculation, etc, review rc-slurm#61 cheaha#44
    • CHEAHA: Refine old modules - ties into rebuild CICD #554 (closed) #566 #480 #460 #432 (closed)
    • CHEAHA: Limit the number of queued OOD jobs per user #673 comment 1
    • DOCS: Migrate docs to GitLab Pages gitlab#7
    • NETWORK: Network upgrade beyond 10G link to campus, beyond 40G to scidmz
    • METRICS: Cheaha job-specific/process-specific observability #661
    • METRICS: Data/Storage observability #661
    • RCS: Services status page #627
    • STORAGE: Globus Azure Blob Connector #685
    • STORAGE: Separate LTS from Core Ceph
    • STORAGE: Shared, read-only research data allocation (e.g. alphafold) #680 #654
    • STORAGE: Standardize /data/project/ #625 #633 #162 (closed)
    • SECURITY: Cheaha observability for security #661
    • SECURITY: RCS Account Statemachine and RabbitMQ refactoring rabbitmq_agents#163
    • SECURITY: RCS Account Database (user reg db) enhancements #393 #400 #82 #410 #141 #642
    • SECURITY: Restrict access to info about users, groups, slurm jobs, etc. cheaha#57
    • TRAINING: Researcher training initiatives
    • UX: RCS main page
    • UX: Expose user reg app content (writing/theme) to facilitation via GitLab account-app#17 #556
    • UX: Expose OOD content (writing/theme) to facilitation via GitLab rc-ood-message#1 #556
    • UX: Expose OpenStack Horizon content (writing/theme) to facilitation via GitLab #624 #556
  • NICE to haves (lower priority)
    • PRODUCTIVITY: Pastebin and etherpad as RC apps
    • CHEAHA: Update Singularity #546 cluster-software#115
    • CHEAHA: Resolve poor researcher UX around pip and conda behaving unexpectedly #485
    • CHEAHA: Internal routing SSH key dedicated storage location
    • DOCS: Docs AI chatbot courtesy of UAB Copilot
    • DOCS: Docs observability and analytics
    • METRICS: Cheaha observability for buy-in model #626 #471
    • METRICS: Cloud.rc observability " #661
    • METRICS: Network observability " #661
    • METRICS: Buffered node exporter data locally for robustness
    • FUNDING: Cheaha compute buy-in model and implementation #626 #637
    • FUNDING: Cloud.rc buy-in model and implementation #672 #637
    • FUNDING: Storage buy-in model and implementation (TiB vs TB is 10%, this matters for dollar costs!) #637
    • STORAGE: Offsite data center backup solution and buy-in model (MUST if funded)
    • STORAGE: Globus Azure Blob connector for backup and archive solution and buy-in model (MUST if funded)
    • STORAGE: Project-parallel scratch directories #653
    • STORAGE: Improve LTS Globus UX to handle multiple keys (individual, lab shared, core shared, ...) #684
    • SECURITY: CMMC compliant enclave
    • PRODUCTIVITY: pastebinit command and self-hosted pastebin equivalent
  • INFOrmational
    • SECURITY: IDM Federation (what is our portion?)

Completed

  • STORAGE: Globus configured for GPFS5 (no issue) #694 (closed)
  • STORAGE: New accounts created on GPFS5
  • GLOBUS: Understand and implement shared administration group and duties in Globus.
  • GITLAB: Migrate data to GPFS5 gitlab#15
  • RCS: RCS now a fully realized platform.
Edited by Fortune Iriaye