rwmjones
commented
2 years ago Closed rwmjones closed 2 years ago
rwmjones
commented
2 years ago rwmjones
commented
2 years ago The yaml config file is:
# Cluster domain
domain: test49.com
# Cluster name
cluster: kata-49
# RHCOS image name in the libvirt storage pool.
# You can leave this empty for kcli to download and use the defaults
# image:
imagecontentsources: []
mdns: True
# Uncomment this line for disconnected install
# disconnected_deploy: True
# Libvirt network name eg. 192.168.10.0/24
network: openshift
# Libvirt storage pool
pool: openshift
# For multiple clusters on the same network it's important to set
# api_ip to unique IP within the network
api_ip: 192.168.10.254
# Copy the pull secret and store it the following file
pull_secret: openshift_pull.json
# Release version number: 4.7/4.8/4.9/4.10
tag: 4.9
# Build type: nightly/stable. The latest nightly or stable build will
# be automatically downloaded
# If specific version is required then download openshift-install from
# https://mirror.openshift.com/pub/openshift-v4/clients and
# place it in /home/kcli/.local/bin:/home/kcli/bin:/usr/local/bin:/usr/bin:/usr/local/sbin:/usr/sbin
# to use the same for install
version: nightly
masters: 3
workers: 2
numcpus: 4
disk_size: 60
# Ensure the memory allocated to the VMs are as per the recommendations.
memory: 16384
bootstrap_memory: 16384
master_memory: 16384
worker_memory: 16384
bootstrap_numcpus: 4
master_numcpus: 4
worker_numcpus: 2
master_macs: []
worker_macs: []
# Uncomment the below lines to install the sandboxed containers operator
# apps:
# - sandboxed-containers-operatorIs test49.com just a made up name? The domain is owned by some Japanese company. Am I supposed to replace that with a local name that will resolve?
karmab
commented
2 years ago better to use a fake domain (or a real one for which you have the correct openshift dns entries in place)
karmab
commented
2 years ago maybe we can check the failure live on your env?
karmab
commented
2 years ago is this actually running on kubevirt?
rwmjones
commented
2 years ago is this actually running on kubevirt?
So I'm not certain what this question means. I'm attempting to get Kubevirt installed in some way and this is all running on a baremetal machine running Fedora 36.
Unfortunately I'm in the middle of preparing to go to a conference in Dublin and won't really be able to look at this in detail until the week after next.
rwmjones
commented
2 years ago So I'm trying again. I'm starting from scratch with a fresh Fedora 36 installation on an Intel NUC. (If there is a different, better tested OS to use, please let me know)
rwmjones
commented
2 years ago I'm using openshift.yaml from https://github.com/bpradipt/infra/blob/master/kcli-yamls/openshift.yaml and the first error was:
Missing rhcos410. Indicate correct image in your parameters file...I thought that kcli could download rhcos images? Anyway I did download it from https://mirror.openshift.com/pub/openshift-v4/x86_64/dependencies/rhcos/4.10/latest/rhcos-qemu.x86_64.qcow2.gz, uncompressed it and placed it into /home/kcli/images/
rwmjones
commented
2 years ago Missing rhcos410. Indicate correct image in your parameters file...
So you have to change image: rhcos410 to image: rhcos-qemu (or rename the image I file I guess). Another case of a non-actionable and non-obvious error.
rwmjones
commented
2 years ago So it later fails with:
$ kcli delete kube --yes testk
Deleting directory /home/kcli/.kcli/clusters/testk
$ kcli create kube openshift --paramfile openshift.yaml
Deploying on client local
Deploying cluster testk
Removing old openshift-install
Using nightly version
Downloading openshift-install from https://mirror.openshift.com/pub/openshift-v4/clients/ocp-dev-preview/latest-4.10
Move downloaded openshift-install somewhere in your PATH if you want to reuse it
Using installer version 4.10.0-0.nightly-2022-01-31-012936
Checking if image rhcos-qemu is available
Deploying disconnected vm testk-disconnected
Deploying Vms...
testk-disconnected skipped on local!
testk-disconnected down
No ip found for testk-disconnected
ssh: Could not resolve hostname none: Name or service not known
ssh: Could not resolve hostname none: Name or service not known
ssh: Could not resolve hostname none: Name or service not known
Setting OPENSHIFT_INSTALL_RELEASE_IMAGE_OVERRIDE to
FATAL failed to fetch Master Machines: failed to load asset "Install Config": invalid "install-config.yaml" file: [imageContentSources[0].mirrors[0]: Invalid value: "/ocp4": failed to parse: invalid reference format, imageContentSources[1].mirrors[0]: Invalid value: "/ocp4": failed to parse: invalid reference format]
Leaving environment for debugging purposes
You can delete it with kcli delete kube --yes testkIt's weird that it's looking for hostname none.
rwmjones
commented
2 years ago I fixed it with Karim's help.