bnode05 reserved for network configuration change in support of RoCE
Scheduled Maintenance Report for CU Boulder RC
Completed
The network configuration change on Blanca 05 has completed successfully, and all nodes have been returned to service.
Posted 6 days ago. Nov 08, 2018 - 11:17 MST
In progress
Scheduled maintenance is currently in progress. We will provide updates as necessary.
Posted 6 days ago. Nov 08, 2018 - 09:15 MST
Scheduled
The Blanca 05 chassis, which contains all nodes with hostname `bnode05*`, is an experiment in providing low-latency MPI via RoCE. However, the switch in this chassis was installed into the wrong slot, which is preventing the RoCE-capable interfaces from operating.

We have reserved the nodes in this chassis via Slurm, and will be reconfiguring their networking in pursuit of enabling RoCE.

Running jobs on Blanca, as well as general Blanca operation, will be unaffected.

https://en.wikipedia.org/wiki/RDMA_over_Converged_Ethernet
Posted 6 days ago. Nov 08, 2018 - 09:15 MST
This scheduled maintenance affected: Blanca.