Red Hat OSP11 Overcloud Trial Deploy Templates for lab on a single Dell R710 using VMware's EVALExperience, ESXi and vCenter
Setup(no vlans)
- 1x Undercloud
- 1x Controller Node
- 1x Computer Node
- 1x Ceph Storage Node
Specs
-
Undercloud: 40GB drive, 4x vCpu, 8GB RAM, 2x nics, external and provisioning
-
Controller: 50GB drive, 4x vCpu, 8GB RAM, 6x nics, provisioning, storageIp, storageMgmt, internalApi, tenant, and external
-
Compute: 200GB drive, 8x vCpu, 16GB RAM, 4x nics, provisioning, storageIp, internalApi, tenant
-
Ceph: 50GB, 200GB, 1TB drives, 4x vCpu, 8GB RAM, 3x nics, provisioning, storageIp, storageMgmt
Networking
I have individual port groups for the names of each service type, provisioning, storageIp, storageMgmt, internalApi, tenant, and external. All but external is mapped to the same vSwitch. Running vyOS as the router for provisioning network, and the external network is directly connected to my home router through a separate vSwitch in VMware. Both the external network, and provisioning network are routable from anywhere in the network.
Challenges
- Adding packages to the overcloud image, like open-vm-tools, and ripgrep [Link]
- Learning to use fake_pxe and power the vmware images on and off at the correct times [Link]
- On PXE boot the interfaces would be correct, however they would all rename themselves during the overcloud deployment causing the deploy to fail since connectivity was lost after the firstboot. I solved this with the user of the interface-mapping.yaml file.
- Mapping the storage drives did not work since VMware drives give no serial, or information about the drive besides the size. I just had to cross my fingers and hope that the 50GB would be recognized as sda, 200GB as sdb, and 1TB as sdc.