Dec 3, 2018 All remaining Newton resources decommissioned (login node/data transfer node, VMs, /data file systems, etc.).
Oct 15, 2018 Newton /lustre file system decommissioned. Users must have transferred remaining files off of Newton /lustre file system
Sept 17, 2018 Newton home directories decommissioned.
Aug 1, 2018 Newton GPFS file system (/gamma) was decomissioned. Any remaining files were tarred and compressed and moved to Newton Lustre file system. See ACF File System documentation.
July 10-11, 2018: ACF maintenance performed to update the JICS 116 Data Center Emergency Power Off (EPO) system The ACF was taken down July 10 at 5pm and returned to service on July 11 at 6pm.
June 2018: Worked with users who had large storage allocations on the Newton /gamma (GPFS) file system to retire this resource from Newton
April 26: 8 additional Skylake nodes added to the ACF - 6 are currently available in production, 2 are reserved for Lustre upgrade testing.
April 23: Added the Skylake with a Volta GPU to ACF in partition "skylake_volta"
March 22: 2-2:30pm Unplanned complete power outage due to TVA operational issue at ORNL. Power to both building 5100 and 5600 lost. UPS on Haven kept Haven storage from having power interrupted. UPS in K200 did not work properly. Infrastructure power interrupted.
March 21: DTNs physically moved from E102 to JICS 116.
March 19: 16 Skylake nodes added to the ACF
March 16: 8:30am to 12:30pm Unplanned Lustre outage due to power issue. Most all I/O recovered when Lustre was restored. Only a few I/O sensitive jobs were interrupted.
March 12 5pm to March 14 5pm: ACF outage to move JICS infrastructure to new location and bring PDU-3 UPS online for storage systems. This was during Spring Break.
March 1, 2018: Significant moab scheduler changes.
February 28, 2018: Changed the sigma partition to split between "sigma" partition for the 24 core nodes and "sigma_bigcore" for the 28 core nodes.
February 21, 2018: Changed the ACF jobs scheduler to increase the reservation depth from 1 to 5. This is the number of jobs that the schedule puts a reservation on specific nodes.
January 31, 2018: 9am to noon: ACF Preventative Maintenance to do software fixes and updates.
January 17, 2018: Medusa file system retired and taken offline. If you need data from Medusa your have 30 days to submit a ticket.
January 5-6, 2018: ACF outage for facilities upgrade.
November 18, 2017: ACF /lustre/medusa will be set to read-only. No new files allowed to be created on /lustre/medusa file systems
November 17, 2017: ACF outage at 5pm due to new boiler water source added to JICS building. System maintenance will be tied in with this outage. Estimated system back online at 5pm Nov 18th
November 1, 2017: Sigma integration complete into ACF. ACF now consists of Beacon, Monster, Rho, Sigma. A few nodes of Rho and Sigma offline due to incomplete hardware.
October 25, 2017: Began Sigma integration into ACF
October 25, 2017: 8am-5pm Scheduled ACF outage due to building chilled water maintenance
October 23, 2017: Newton /lustre and /gamma file systems configured as read-only. No new files allowed to be created on these file systems
October 6-9, 2017: Newton offline due to power outage in KPB. Network access to ACF resources at JICS will be affected from campus and most all external locations
October 4-6, 2017: The Newton Sigma nodes no longer available for user jobs and are taken offline to be moved to JICS
September 14: Users begin migrating Newton files and ACF Lustre Medusa files to ACF Lustre Haven file system
September 13, 2017: New Lustre Haven file system scheduled for production use in the ACF at end of day
September 11, 2017: Rho nodes integrated into ACF
August 14, 2017: Newton Rho nodes are taken offline and moved to JICS
August 11, 2017: Monster node integrated into ACF
July 28, 2017: One Rho chassis and Sigma chassis taken offline from Newton and moved to JICS
July 26, 2017: ACF began with Beacon partition, application for accounts opens
July 3-25, 2017: Beacon down for OS upgrades and conversion to ACF
July 7, 2017: Newton Monster node was taken offline