crw-pluginregistry 2.7 job has been failing consistently all week while running the rhel.install.sh script to install rpms, including yq and skopeo, which are required for the rest of the build.

See:

https://main-jenkins-csb-crwqe.apps.ocp4.prod.psi.redhat.com/job/CRW_CI/job/crw-pluginregistry_2.7/23/console (and previous builds too)

Oddly, the devfile registry does NOT have this problem, so perhaps it's worth refactoring the plugin reg build to do what the devfile reg build does when installing rpm deps.

But these look identical...

• https://github.com/redhat-developer/codeready-workspaces/blob/crw-2.7-rhel-8/dependencies/che-devfile-registry/build/dockerfiles/rhel.install.sh
• https://github.com/redhat-developer/codeready-workspaces/blob/crw-2.7-rhel-8/dependencies/che-plugin-registry/build/dockerfiles/rhel.install.sh

as do these:

• https://github.com/redhat-developer/codeready-workspaces/tree/crw-2.7-rhel-8/dependencies/che-devfile-registry/build/dockerfiles
• https://github.com/redhat-developer/codeready-workspaces/tree/crw-2.7-rhel-8/dependencies/che-plugin-registry/build/dockerfiles

So... perhaps the Jenkinsfiles set up different environment / global rules in dnf.conf?

Or else the changes that dnochevn has been doing in https://github.com/eclipse/che/pull/19103/ (and the suggestions I've made there) are the issue here?

Since the specific rpms that cause the install to fail vary from build to build, it seems like increasing pulp repo timeouts / retries might help.

Jobs are in:

https://gitlab.cee.redhat.com/codeready-workspaces/crw-jenkins/-/tree/master/jobs/CRW_CI

This is urgent, needs to be fixed this week so we can update to 7.26.2, respin, and be done for code freeze by end of the week.