Setup/config OpenShift test-environment

Posted on

Hi,

Red Hat OCP is new for me.

In order to gain experience with the product I installed a test OpenShift 4.16 cluster of 3 nodes (master/worker combined) using the assited installation trough the Hybrid Cloud Console with virtualization enabled, so with a 90 days trial subscription!
As a result I have following questions regarding the assited installation and usage:

1) Network-setup cluster-nodes:
The installation has been done using only 1 1GB interface as I could not find a way to configure the network with a bond interface using 4 1GB interfaces in LACP?
- Is it possible to configure this during the assisted installation, and if so how?
- Can we reconfigure the cluster-nodes easily to make use of the bond-interface, and if so how?

2) We also make use of a Service-server (DNS/HAproxy).
In the HAproxy status overview only 2 of the 3 nodes are UP for the ports 80/443.
Is this normal behavior or should I have the 3 nodes having the status UP?

3) As persistent storage we would make use of an iSCSI disk on Synology, connected by a LACP bond-interface of 4 1GB interfaces.
I could not find the possibility to configure this iSCSI device during the assisted installation process.
If this should be possibel, I would like to know how.
So I configured it manually after installation:
- created a default storageClass using the provisioner "kubernetes.io/iscsi" with the necessary config of the iSCSI device
- modified the storageprofile (same name as the created storageclass) to use both accessModes RWO and RWX.
This results in Openshift that creates PVC's for os-images it would import. But I have to create PV's in order to be bound and start.
Once done the import starts and the download goes fast.
Can the creation of the PV's be automated once a PVC is created?

4) VM creation:
As mentioned above for the os-imports, a PVC for a test VM (fedora) is also created automatically, but I had to create a PV manually in order to bound it with the PVC before the the vm-installation continued.
So also here, is it possible that the PV would be created automatically?
Once done it continues with the cloneprocess of the image.
But this goes extremely slow, it took more than 2 days!
The slow cloneprocess might be caused by the missing LACP bond interface on the cluster-node, so a solution to reconfigure the network for cluster-nodes would be appreciated.

Thanks in advance for any feedback/help.

Regards,

Geert

GW

Responses