ROSAENG-61841: Prow e2e migration - boilerplate update and fix DaemonSet->Deployment in e2e

boilerplate/_data/backing-image-tag

@@ -1 +1 @@
-image-v8.3.6
+image-v8.4.3

boilerplate/_data/last-boilerplate-commit

@@ -1 +1 @@
-8fb7c801f68dc7e06e8d2ae138c2a98f0b234b56
+a0e42e58ed1d65bb75a848c595b34ae5553296eb

boilerplate/_lib/subscriber-propose-update

@@ -25,7 +25,7 @@ Quirks and Limitations:
 - Is still slightly interactive, because 'gh pr create' likes to ask
   questions about your origin and upstream.
 EOF
-    exit -1
+    exit 1
 }
 
 source $REPO_ROOT/boilerplate/_lib/subscriber.sh
@@ -34,47 +34,101 @@ source $REPO_ROOT/boilerplate/_lib/subscriber.sh
 [[ $# -eq 0 ]] && usage
 
 TMPD=$(mktemp -d)
+echo $TMPD;
 trap "rm -fr $TMPD" EXIT
 
+run_step() {
+    local title=$1
+    local log_file="$TMPD/$title.log"
+    log_file=$(tr '[:upper:]' '[:lower:]' <<< "$log_file")
+    log_file=$(tr ' ' '-' <<< "$log_file")
+    shift
+
+    if [[ $1 != "--" ]]; then
+        echo "ERR: expected '--' but got '$1'"
+        exit 1
+    fi
+    shift
+    echo -n "$title...  "
+
+    if ! "$@" > "$log_file" 2>&1; then
+        echo " FAILED"
+        echo "!!!"
+        echo "!!! Boilerplate update failed for $subscriber"
+        echo "!!!"
+        echo ""
+        cat "$log_file"
+        exit 1
+    fi
+    echo " DONE"
+}
+
+sync_main() {
+    local main_branch=$1
+    shift
+
+    git pull upstream $main_branch
+    git push origin $main_branch
+}
+
+git_clean_and_push() {
+    local branch=$1
+    shift
+
+    git push --delete origin $branch || true
+    git push -u origin $branch
+}
+
 propose_update() {
     local subscriber=$1
     local proj=${subscriber#*/}
 
-    if [[ -z "$DRY_RUN" ]]; then
-        echo "DRY RUN: Would propose update for $subscriber"
-        return 0
-    fi
-
     (
         # Clone my fork of the subscriber repo
         cd $TMPD
         # This
         # - uses the existing fork if one exists
         # - sets 'origin' and 'upstream' remotes
-        gh repo fork $subscriber --clone=true --remote=true
+        # only clones the default branch to save disk space and time
+
+        run_step "Creating fork" -- gh repo fork $subscriber --clone=true --default-branch-only
         cd $proj
 
-        # Current branch is 'master' or 'main'
-        cur_branch=$(current_branch .)
-        # Make sure our origin is synced with upstream, so our update
-        # commit is based off of the latest code.
-        # WARNING: This changes your fork!
-        git pull upstream $cur_branch
-        git push origin $cur_branch
-
-        # Create the update commit
-        make boilerplate-update
-        make boilerplate-commit
-
-        # And create the PR
-        # TODO: This is interactive. How do we tell gh "Yes, please use
-        # upstream as upstream and origin as origin?"
-        gh pr create -f
+        # Current branch is 'master' or 'main' or 'trunk'
+        main_branch=$(current_branch .)
+        run_step "Syncing Fork" -- sync_main $main_branch
+        # run_step "Pushing fork" -- git push origin $main_branch
+
+        # Create the update commit - only cat logs if something goes wrong.
+        run_step "Updating boilerplate" -- make boilerplate-update
+        run_step "Committing boilerplate update" -- make boilerplate-commit
+
+        boilerplate_branch=$(git rev-parse --abbrev-ref HEAD)
+        # By pushing to the origin boilerplate branch explicitly before opening a PR,
+        # we make don't get prompted for the branch to push to.
+        # If we still find that it's giving us an interactive prompt, we can otherwise
+        # use `gh api` to create the PR programmatically.
+        if [[ "$boilerplate_branch" == "$main_branch" ]]; then
+            echo "CRITICAL ERROR: boilerplate branch '$boilerplate_branch' is the same as main branch '$main_branch'"
+            echo "If you see this, something has gone terribly wrong"
+            echo "Skipping"
+            exit 20
+        fi
+        run_step "pushing update" -- git_clean_and_push $boilerplate_branch
+
+        gh pr create --repo $subscriber -f $DRY_RUN_FLAG
     )
 }
 
 bp_master=$(git rev-parse master)
 
+DRY_RUN_FLAG=""
+if [[ -z "$DRY_RUN" ]]; then
+    echo "DRY RUN: ENABLED"
+    DRY_RUN_FLAG="--dry-run"
+fi
+
+
 for subscriber in $(subscriber_args "$@"); do
 
     # Does this one need an update?
@@ -89,14 +143,45 @@ for subscriber in $(subscriber_args "$@"); do
         continue
     fi
 
-    # Is there already a PR proposed for this level?
-    existing_pr=$(gh pr list --repo $subscriber | grep -P ":boilerplate-\S+-$bp_master\s")
+    # Is there already a PR proposed for this commit?
+    pr_list=$(gh pr list --repo $subscriber --json headRefName,url,number | jq -r '. | map(select(.headRefName | startswith("boilerplate-update--")))')
+    existing_pr=$(jq -r ".[] | select(.headRefName == \"boilerplate-update--$bp_master\")" <<< "$pr_list")
     if [[ -n "$existing_pr" ]]; then
-        echo "Subscriber '$subscriber' already has an open PR:"
-        echo "https://github.com/$subscriber/pull/$existing_pr"
+        echo "Subscriber '$subscriber' already has an open PR for this boilerplate commit:"
+        jq -r .url <<< "$existing_pr"
         continue
     fi
 
     # Pull the trigger
-    propose_update "$subscriber"
+    if ! propose_update "$subscriber"; then
+        echo "Error: failed to propose update for '$subscriber'"
+        continue
+    fi
+
+    new_pr="XXXX"
+    # Get the new PR URL
+    # only run if not dry-run - otherwise the new_pr var will be empty
+    if [[ -n $DRY_RUN ]]; then
+        new_pr=$(gh pr list --repo $subscriber --json headRefName,number | jq -r ".[] | select(.headRefName == \"boilerplate-update--$bp_master\") | .number")
+        if [[ -z "$new_pr" ]]; then
+            echo "error: unable to find new PR for boilerplate update '$bp_master' on subscriber '$subscriber'"
+            continue
+        fi
+    fi
+
+    # Add comments to existing PRs to say they're superseded by this new one
+    if [[ -n "$pr_list" ]]; then
+        prs=$(jq -r '. | map(.number) | @tsv' <<< "$pr_list")
+        echo "Closing old PRs: $prs"
+        for pr in $prs; do
+            if [[ -z $DRY_RUN ]]; then
+                echo "Dry run - would close $pr with comment:"
+                echo "  \"Superseded by #$new_pr.\""
+                continue
+            fi
+
+            gh pr close --repo $subscriber --comment "Superseded by #$new_pr." $pr
+        done
+    fi
+
 done

boilerplate/openshift/golang-osd-e2e/OWNERS

@@ -1,4 +1,4 @@
 reviewers:
-- srep-infra-cicd
+- rosa-staff-engineers
 approvers:
-- srep-infra-cicd
+- rosa-staff-engineers

boilerplate/openshift/golang-osd-e2e/README.md

@@ -31,8 +31,48 @@ following:
 |------------------------|---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|
 | `e2e-binary-build`     | Compiles ginkgo tests under test/e2e and creates the ginkgo binary.                                                                                                                                                                                                                                                                                                                                                                                                                               |
 | `e2e-image-build-push` | Builds e2e image and pushes to operator's quay repo. Image name is defaulted to <operator-image-name>-test-harness. Quay repository must be created beforehand.                                                                                                                                                                                                                                                                                                                        |
+| `e2e-local`            | Builds the e2e binary and runs it against a cluster via KUBECONFIG or backplane. Supports focused tests via GINKGO_FOCUS.                                                                                                                                                                                                                                                                                                                                                              |
 
 #### E2E Local Testing
 
-Please follow [this README](https://github.com/openshift/ops-sop/blob/master/v4/howto/osde2e/operator-test-harnesses.md#using-ginkgo) to run your e2e tests locally
+Run e2e tests locally against a managed cluster without waiting for the full Prow CI pipeline.
 
+**Prerequisites:**
+- `ocm` CLI logged into the appropriate environment (`ocm login --use-auth-code --url staging`)
+- Access to a managed cluster (via KUBECONFIG or backplane)
+- Go toolchain installed
+
+**Option 1: Using backplane (recommended)**
+
+```bash
+# Run all tests against a cluster by ID
+make e2e-local CLUSTER_ID=2rmlgv5dbdp2285n85o7h3aaa6pafkpq
+
+# Run focused tests
+make e2e-local CLUSTER_ID=2rmlgv5dbdp2285n85o7h3aaa6pafkpq GINKGO_FOCUS="is installed"
+
+# Run tests with a label filter
+make e2e-local CLUSTER_ID=2rmlgv5dbdp2285n85o7h3aaa6pafkpq GINKGO_LABEL_FILTER="!slow"
+```
+
+**Option 2: Using an existing KUBECONFIG**
+
+```bash
+# Set KUBECONFIG to your cluster's kubeconfig
+export KUBECONFIG=/path/to/kubeconfig
+
+# Run all tests
+make e2e-local
+
+# Run a single test
+make e2e-local GINKGO_FOCUS="reconciles required resources"
+```
+
+**Output:**
+- Test results print to stdout with verbose Ginkgo output
+- JUnit XML report saved to `e2e-local-junit.xml`
+
+**Tips:**
+- Use lease clusters for testing (check `#rosa-prow-info` for available clusters)
+- The operator must be deployed on the target cluster (via PKO or OLM)
+- If tests fail with "not found" errors, verify the operator is running: `oc get deployment -n <operator-namespace>`