wiki:2024-9-5

Version 2 (modified by Tom Goddard, 14 months ago) ( diff )

--

Attendees

Tom Goddard, Piet, Scooter, Greg, Eric, Elaine

September 5, 2024

Agenda

  • Scooter work in progress, TODO: Add link to slides

Discussion Notes

  • Wynton downtime
    • Probably will have a 5 day Wynton down time the week of October 14 to run fsck file system check.
  • Endowed chair
    • Search committee to hire for RBVI endowed chair is being formed, Tom Ferrin is committee chair.
    • Letter written to get Chancellor's approval for endowed chair. Describes position as bioviz and informatics and software development.
  • Plato cluster local disk
    • We have ordered a networke attached storage (NAS) device for 4 plato nodes to use for web hosting, web services, Trac, mail and other /usr/local stuff.
    • The hope is that moving these off the Wynton beegfs file system will avoid downtimes effecting these services.
    • Synology Rackstation RS3621xs+1, 12 bays
    • Ordered three 4 TByte ssd drives that arrived today and will use 4 existing SSD drives from plato nodes to create Raid 6 array plus one spare drive, 20 Tbytes storage
    • Extra 5 drives can be purchased for snapshots
    • NAS has 2 10 Gbit/sec ethernet connections.
    • Scooter will try to use 2 x 10Gbit/s bonded but probably our switch won't handle that with copper. So maybe will use one 10 Gbit/s each to 2 nodes.
    • Will try to increase the 8 GB of cache memory with old plato memory (1.2 TB available).
    • Can also use two NVME slots for cache on NAS.
    • Will mount by NFS with caching on NAS and client nodes.
    • NAS file system is btrfs, we are not familiar with.
    • Scooter and Greg will test and migrate databases, mailman, web server, web services, trac in November to December time frame.
    • Home directories stay on Wynton beegfs for simplicity.
  • Wynton audit
    • Scooter is technical lead on wynton audit
    • Working on sending wynton logs to UCSF IT encrypted.
    • Working on DUO 2-factor for all wynton admin logins.
    • Working on data loss prevention plan.
    • Home directory backups, currently daily snapshots for 2 weeks, only for home directories under 5 million files.
    • Using rclone for backups.
    • No backups of lab group directories, funding fell through.
  • SPOKE
    • Configured neo4j database to use more memory on plato.

Action Items

Note: See TracWiki for help on using the wiki.