DFSc Working Group

Meeting Date

• TEAMS: 2024-08-08

Invitees

• Gwen, Guoxiang, Devon, Clayton, Lori, Nathan

Attendees

Agenda Items

State of DFSc

• Still running Ceph 16
• Containerized Ceph deployment makes monitoring and investigating problems difficult, deploying targetted fixes very difficult
• Ongoing Kerberized NFS issues
• Fragility
  • Failure of any of 9(soon 12) servers will result in a persistent degraded state
• Lack of snapshots
• Poor resource utilization
  • Storing database backups

Upcoming Work

• Re-enable snapshots
  • Create DFSc admin node RT: 11319804
  • Implement hourly snapshots RT: 1319805
  • Clean up old snapshots RT: 1319806
  • Define retention policy and implement retention management RT: 1319807

• Move database backups
  • Postgres backups RT: 1319808
  • MySQL backups RT: 1319809

• Hardware deployment and incremental upgrades
  • Test upgrading cephadm managed Ceph 16 on Ubuntu 20.04 to native Ceph 17 packages on Ubuntu 22.04 using *-902 1319810
  • Upgrade production cluster to Ceph 17 on Ubuntu 22.04
  • Test Ceph 17 on *-423 hardware RT: 1319803
    • Performance tuning
  • Integrate *-423 hardware into production cluster
  • Migrate (some) TEACHING homedir data
  • Upgrade production cluster to Ceph 18 on Ubuntu 24.04

• Procure more drives for *-423 hardware RT: 1319708
  • Migrate remaining homedir data

• Ongoing Ganesha bugfixing
  • Solid lead on most critical load sensitive bug RT: 1315069