openshift/ci-docs

Openshift CI docs

This repository holds the content of the Openshift CI documentation at https://docs.ci.openshift.org/

Development

1. Clone the repository including submodules: git clone --recursive [email protected]:openshift/ci-docs.git. NOTE: Git submodules must be fully cloned for the server to work. If you've already cloned the repository without submodules, run git submodule update --init --recursive --depth 1.
2. Install the extended version of hugo as described here. Generally, installing the extended binary from their releases page is the best option.
3. Start the hugo server via hugo server
4. Edit content and watch your changes live at http://localhost:1313/

ci-docs's Issues

make new-repo and CONTAINER_ENGINE

It may be worth adding to the documentation that the CONTAINER_ENGINE environment variable needs to be set when using podman.
I have an alias for docker redirecting to podman, running make new-repo led to a "permission denied" error. When running the make command the container is executed with the --user option. Adding --userns=keep-id avoids the "permission denied" issue with podman. Alternatively not having --user (the original intent in the Makefile when CONTAINER_ENGINE is set) works well too.
After the failure the user gets instructured to rerun:
/usr/bin/repo-init --release-repo /release --config="{..}"
which is not correct since this command is now run inside a container. This may be better to point to something like:
podman run --platform linux/amd64 --rm -it -v "/home/fgiloux/development/release:/release:z" registry.ci.openshift.org/ci/repo-init:latest --release-repo /release --config="{...}"

path to pr-reminder in incorrect

should be
The configuration for pr-reminder is located in the openshift/release repo under clusters/app.ci/pr-reminder.

make new-repo requires Python dependencies to be installed

CI docs imply that make new-repo can be run by itself, no prerequisites necessary [1]. This is not true - at some point make boskos-config is run, which requires the yaml Python module.

Either:

1. Docs should be updated so users install the yaml module in their environment.
2. make boskos-config is updated to run within a container via docker/podman.

[1] https://docs.ci.openshift.org/docs/how-tos/onboarding-a-new-component/#bootstrapping-configuration-for-a-new-repository

some links are difficult to navigate on docs landing page

some of the links on the main docs page, most notably the "Contributing CI Configuration to the openshift/release Repository" and "Using External Images in CI", are difficult to navigate due to some sort of padding or overlap that is happening with the header elements below them.

this image shows the issue more clearly

Testing Operators Built With The Operator SDK and Deployed Through OLM

In the Building Operator Bundles section, it is not clear from the docs that dockerfile_path is relative to context_dir.

The example given in the docs (see below) suggests that it is not relative because path/ is present at the beginning of both.

operator:
 bundles: # entries create bundle images from Dockerfiles and an index containing all bundles
 - as: my-bundle                         # optional
   dockerfile_path: "path/to/Dockerfile" # defaults to `bundle.Dockerfile`
   context_dir: "path/"                  # defaults to .

Wrong Variables referenced in Testing Operators Built With The Operator SDK and Deployed Through OLM

The section Operator Upgrade Testing has a typo regarding the initial channel and CSV variables. According to the optional-operators-subscribe step the variables should are named as:

• OO_INITIAL_CHANNEL (docs) -> INITIAL_CHANNEL (step)
• OO_INITIAL_CSV (docs) -> INITIAL_CSV (step)

Adding a New Secret to CI - Creating the manifests

In the "Creating the manifests" section, the docs refer to a core-services/secrets directory that does not exist in openshift/release.

Interacting With CI Image Registries: Flag --service-account has been deprecated

In https://docs.ci.openshift.org/docs/how-tos/use-registries-in-build-farm/#how-do-i-get-a-token-for-programmatic-access-to-the-central-ci-registry, I now get:

$ oc --namespace rhcos-devel registry login --service-account image-pusher --registry-config=ocp-ci-registry-push.json
Flag --service-account has been deprecated, and will be removed in the future version. Use oc create token instead.

Note: It's still working for me right now.

I've tried:

$ oc --namespace rhcos-devel registry login --service-account image-pusher --registry-config=ocp-ci-registry-push.json
error: failed to create token: serviceaccounts "image-pusher" is forbidden: User "travier" cannot create resource "serviceaccounts/token" in API group "" in the namespace "rhcos-devel"

best_effort requires a timeout, and I can't find docs for which timeout

I have added best_effort: true to a ref as per the docs, however the config presubmit fails because

failed to load registry: test openstack-gather contains best_effort without timeout

I believe this is not documented?

cf: https://prow.ci.openshift.org/view/gs/origin-ci-test/pr-logs/pull/openshift_release/18989/pull-ci-openshift-release-master-ci-operator-config/1400388560363720704

Have search results show how/where the search term matches

The results do not show where in the page the term is matched which makes it hard to consider which match is the most relevant. e.g. search for OWNERS.

javascript error expanding existing cluster pools

Javascript error expanding cluster pool status

https://docs.ci.openshift.org/docs/how-tos/cluster-claim/#existing-cluster-pools

:formatted:56 Uncaught TypeError: Cannot read properties of undefined (reading 'releaseImage')
    at format (:formatted:56)
    at HTMLTableCellElement.<anonymous> (:formatted:94)
    at HTMLTableSectionElement.dispatch (jquery-3.5.1.min.js:2)
    at HTMLTableSectionElement.v.handle (jquery-3.5.1.min.js:2)
format @ :formatted:56
(anonymous) @ :formatted:94
dispatch @ jquery-3.5.1.min.js:2
v.handle @ jquery-3.5.1.min.js:2

Seems the d argument passed to format() is nil

Set Up Slack Alerts for Periodic Job Results

Update Slack workspace name, from 'coreos' to 'redhat-internal'

Observer Pods

It seems that any type of observer pod that is being added needs to be in 'observers' folder under step-registry. Otherwise I got the below error

Pr reference: openshift/release#39697

time="2023-05-24T22:09:39Z" level=error error="failed to validate configuration ci-operator/config/redhat-chaos/krkn-hub/redhat-chaos-krkn-hub-main.yaml: Failed resolve MultiStageTestConfiguration: observer \"redhat-chaos-cerberus-observer\" is referenced but no such observer is configured"

time="[2](https://prow.ci.openshift.org/view/gs/origin-ci-test/pr-logs/pull/openshift_release/39697/pull-ci-openshift-release-master-ci-operator-config/1661493943134588928#1:build-log.txt%3A2)023-05-24T22:09:39Z" level=fatal msg="error validating configuration files"

Search results are linked to the scroll position of page

When using the search box, the results are returned in a window that appears to be linked to the scroll position on the page.

For example, in this first picture i have used the search box to search while the page is scrolled to the middle of the doc content. you can see that my search results extend up past the top of the screen:

In this second picture, i have only scrolled the page using my mouse wheel and it shows more of the search results:

Ideally, the search results would pin to the location of the input form, or maybe the nav window.