Frequently Asked Questions (FAQ)¶
Below are answers to common questions regarding OKD installation and administration. If you have a suggested question or a suggested improvement to an answer, please feel free to reach out.
What are the relations with OCP project? Is OKD4 an upstream of OCP?¶
In 3.x release time OKD was used as an upstream project for Openshift Container Platform. OKD could be installed on Fedora/CentOS/RHEL and used CentOS based images to install the cluster. OCP, however, could be installed only on RHEL and its images were rebuilt to be RHEL-based.
Universal Base Image project has enabled us to run RHEL-based images on any platform, so the full image rebuild is no longer necessary, allowing OKD4 project to reuse most images from OCP4. There is another critical part of OCP - Red Hat Enterprise Linux CoreOS. Although RHCOS is an open source project (much like RHEL8) it's not a community-driven project. As a result, OKD workgroup has made a decision to use Fedora CoreOS - open source and community-driven project - as a base for OKD4. This decision allows end-users to modify all parts of the cluster using prepared instructions.
It should be noted that OKD4 is being automatically built from OCP4 ci stream, so most of the tests are happening in OCP CI and being mirrored to OKD. As a result, OKD4 CI doesn't have to run a lot of tests to ensure the release is valid.
These relationships are more complex than "upstream/downstream", so we use "sibling distributions" to describe its state.
How stable is OKD4?¶
OKD4 builds are being automatically tested by release-controller. Release is rejected if either installation, upgrade from previous version or conformance test fails. Test results determine the upgrade graph, so for instance, if upgrade tests passed for beta5->rc edge, clusters on beta5 can be directly updated to rc release, bypassing beta6.
The OKD stable version is released bi-weekly, following Fedora CoreOS schedule, client tools are uploaded to Github and images are mirrored to Quay.
Can I run a single node cluster?¶
Currently, single-node cluster installations cannot be deployed directly by the 4.7 installer. This is a known issue. Single-node cluster installations do work with the 4.8 nightly installer builds.
As an alternative, if OKD version 4.7 is needed, you may have luck with Charro Gruver's OKD 4 Single Node Cluster instructions. You can also use Code Ready Containers (CRC) to run a single-node cluster on your desktop.
What to do in case of errors?¶
If you experience problems during installation you must collect the bootstrap log bundle, see instructions
If you experience problems post installation, collect data of your cluster with:
oc adm must-gather
See documentation for more information.
Upload it to a file hoster and send the link to the developers (Slack channel, ...)
During installation the SSH key is required. It can be used to SSH onto the nodes later on -
ssh core@<node ip>
Where do I seek support?¶
OKD is a community-supported distribution, Red Hat does not provide commercial support of OKD installations.
Contact us on Slack:
- Workspace: Kubernetes, Channel: #openshift-users (for installation & usage questions/issues)
For OKD-specific help, visit the community-led discussion https://github.com/okd-project/okd/discussions forum
See https://openshift.tips/ for useful Openshift tips
Where can I find upgrades?¶
Nightly builds (from
4.x.0-0.okd) are pruned every 72 hours.
If your cluster uses these images, consider mirroring these files to a local registry.
Builds from the stable-4 stream are not removed.
How can I upgrade my cluster to a new version?¶
Find a version where a tested upgrade path is available from your version for on
- Web Console: Home -> Overview -> Tab: Cluster, Card: Overview -> View settings -> Update Status
- Shell: Upgrades to latest available version
oc adm upgrade
Upgrade to a certain version (will ignore the update graph!)
oc adm upgrade --force --allow-explicit-upgrade=true --to-image=registry.ci.openshift.org/origin/release:4.4.0-0.okd-2020-03-16-105308
This will take a while; the upgrade may take several hours. Throughout the upgrade, kubernetes API would still be accessible and user workloads would be evicted and rescheduled as nodes are updated.
Interesting commands while an upgrade runs¶
Check overall upgrade status:
oc get clusterversion
Check the status of your cluster operators:
oc get co
Check the status of your nodes (cluster upgrades may include base OS updates):
oc get nodes
How can I find out what's inside of a (CI) release and which commit id each component has?¶
This one is very helpful if you want to know if a certain commit has landed in your current version:
oc adm release info registry.ci.openshift.org/origin/release:4.4 --commit-urls
Pull From: registry.ci.openshift.org/origin/release@sha256:79b82f237aad0c38b5cdaf386ce893ff86060a476a39a067b5178bb6451e713c
machine-os 31.20200407.20 Fedora CoreOS
How can I find out the version of a particular package within an OKD release?¶
# Download and enter the machine-os-content container.
podman run --rm -ti `oc adm release info quay.io/openshift/okd:4.13.0-0.okd-2023-06-24-145750 --image-for=machine-os-content`
# Query the particular rpm. For example, to get the version of the cri-o package in the release, use the following:
rpm -qa cri-o
How to use the official installation container?¶
The official installer container is part of every release.
# Find out the installer image.
oc adm release info quay.io/openshift/okd:4.7.0-0.okd-2021-04-24-103438 --image-for=installer
# Example output
# Run the container via Podman or Docker to perform tasks. e.g. create ignition configurations
docker run -v $(pwd):/output -ti quay.io/openshift/okd-content@sha256:521cd3ac7d826749a085418f753f1f909579e1aedfda704dca939c5ea7e5b105 create ignition-configs