Open dledford opened 11 months ago
Lincoln suggested we simply create an ansible directory at the top of https://github.com/OpenFabrics/fsdp_setup
instead of making an entirely new directory. This is mainly because there isn't a technical reason for a 2nd repo as opposed to a simple directory, working solely within fsdp_setup allows us to remove/add components from the bash script and to the ansible setup script in 1 PR, and because this repo is already automatically checked out to the cluster hosts.
Does this sound okay?
Per the network manager team:
Device renaming requires udev rules (NM doesn't do any renaming)
Creating persistent state can be done in one of two ways:
Drop a yaml file in /etc/nmstate and enable the nmstate.service
Wipe out and create new files in /etc/NetworkManager/system-connections/ to represent the devices
RHEL 7 isn't supported by nmstate so we are going to stick with using the systemd method of setting up the network interfaces. Once we drop support for RHEL 7, we can consider using nmstate to set up the network interfaces.
I think we can safely drop rhel7 support.
When adding the NVMe drives, we changed what cards are installed in nodes 01 and 02, and also removed the bifurcating PCI-e card from the Mellanox cards in nodes 09 and 10. We need to update the machine definition files in the fsdp_setup scripts to compensate.