Skip to content

Home

Jump to bottom
Tim Walsh edited this page Oct 30, 2020 · 4 revisions

Important

Ensure firewall has https service allowed before running expose_cluster.sh.

Once the script is run and client machines have the ocp4 cluster address registered in /etc/hosts on Linux or C:\Windows\System32\drivers\etc\hosts the ocp cluster can be accessed from a browser remotely from https://console-openshift-console.apps.ocp4.local

If can't connect check

  1. Firewall had https service open before running the expose_cluster.sh script. Rerunning the script will indicated which rules to delete so it can be re-run successfully.
  2. Ensure the expose_cluster.sh script is run.
  3. Ensure hosts files on remote machines have ip address of cluster server.

Performance

Best results are obtained when using NVME or SSD storage.

There is high i/o and a drive with high latency causes many errors to occur and even booting cluster to take 12 hours and in some cases not boot at all.

Running iostat will show the i/o load.

Centos 8

The default installation has DNS on port 53.

To use dnsmasq instead of NetworkManager for DNS

  1. /etc/NetworkManager/NetworkManager.conf line dns=none is added.
[main]
dns=none
  1. Restart NetworkManager
systemctl restart NetworkManager
  1. /etc/resolv.conf
nameserver 127.0.0.1
search modem localdomain
nameserver 192.168.0.1
nameserver 2001:8003:7841:3000::1

Fedora 32

Instructions on Following Setting up dnsmasq on the Hypervisor work fine for Fedora 32.

Ensure firewall has https service allowed before running expose_cluster.sh

Clone this wiki locally