This video continues on from the first one, where we migrated workloads from oVirt, to Fedora Server. In this video, we deploy OKD using the User Provisioned Infrastructure method, discuss the bootstrap process and troubleshoot a couple of issues related to the Machine Config Operator and HAProxy.
Install guide for UPI:
[ Ссылка ]
Machine Config Operator issue that I hit:
[ Ссылка ]
[ Ссылка ]
Useful commands:
Download the ISO related to your openshift-installer version:
curl -O $(./openshift-install coreos print-stream-json | jq -r '.architectures.x86_64.artifacts.metal.formats.iso.disk.location')
Compare the in-cloud Machine Config, compared to what is running on the nodes:
in-cluster version:
oc get mc -ojson rendered-master-cde381fd33e9ab75849d84c6e896664d
Version running on nodes:
ssh core@172.20.0.40 cat /etc/mcs-machine-config-content.json
The diff between the two will reveal what the problem is in the cluster.
0:00 Intro Recap Part1
0:30 VLAN interfaces libvirt XML
1:50 Which version of Fedora CoreOS
4:00 Bootstrap VM
6:52 DHCP settings
8:16 install-config.yaml
9:20 Generate manifests and ignition
10:15 ignition files to webserver
11:20 Boot VMs from FCOS ISO
12:25 coreos-installer command
12:40 Login to bootstrap
15:06 ctlplane nodes
17:37 ctlplane looking for Machine Config
19:15 wait-for bootstrap-complete
20:20 API up
21:42 HAProxy issue and explanation of SSL cert issue
23:13 Adding workers
27:20 Adding DHCP entry
29:39 CSR signing for workers
33:04 Check pods on nodes
35:00 Explanation of Machine Config issues
