ci: Add support for configuring the global catalog namespace #2825
Conversation
openshift-ci
bot
added
the
approved
Signed-off-by: timflannagan <timflannagan@gmail.com>
timflannagan
force-pushed
the
ci/configure-global-catalog-namespace
branch
from
37042cc to
8566c55
Compare
| endif | ||
| E2E_OPTS ?= $(if $(E2E_SEED),-seed '$(E2E_SEED)') $(if $(SKIP), -skip '$(SKIP)') $(if $(TEST),-focus '$(TEST)') $(if $(ARTIFACT_DIR), -output-dir $(ARTIFACT_DIR) -junit-report junit_e2e.xml) -flake-attempts $(E2E_FLAKE_ATTEMPTS) -nodes $(E2E_NODES) -timeout $(E2E_TIMEOUT) -v -randomize-suites -race -trace -progress | ||
| E2E_INSTALL_NS ?= operator-lifecycle-manager | ||
| E2E_CATALOG_NS ?= $(E2E_INSTALL_NS) |
There was a problem hiding this comment.
Setting this to the same value as the OLM installation namespace is intention. This seemed like the best approach to reduce the amount of variables that needed to be overridden between the e2e and e2e-local targets. For upstream environments, you may only need to run make e2e E2E_CATALOG_NS=olm for clusters that were already created, and for the e2e-local target, it should be a no-op but there's no integration tests that can easily test all these configuration permutations.
|
Note: this should fix a permafailing downstream OLM testing flake. See openshift/operator-framework-olm#345 which was a sanity check before opening these changes. When running the OLM testing suite on an OCP (or crc) cluster, you could run the following: |
|
[APPROVALNOTIFIER] This PR is APPROVED This pull-request has been approved by: dinhxuanvu, timflannagan The full list of commands accepted by this bot can be found here. The pull request process is described here
Needs approval from an approver in each of these files:
Approvers can indicate their approval by writing |
|
/lgtm |
Signed-off-by: timflannagan timflannagan@gmail.com
Description of the change:
Add support for configuring the global catalog namespace in the e2e testing suite. Introduces a CLI flag for configuring the global catalog namespace for OLM. This kind of configuration is already present in the catalog-operator's CLI flags, so this is essentially a no-op for upstream OLM given we deploy the OLM stack and the catalog sources in the same namespace. These changes should make it easier for different distributions of OLM, e.g. openshift, that deploy any default catalog source resources in a dedicated namespace.
Motivation for the change:
Accommodate different distributions of OLM that have dedicated catalog namespaces. Reduce any testing flakes for OLM.
Architectural changes:
N/A.
Testing remarks:
Reviewer Checklist
/doc[FLAKE]are truly flaky and have an issue