WIP: ci-operator/step-registry/ipi/conf/workload: Synthetic workload step

ci-operator/step-registry/ipi/conf/workload/OWNERS

@@ -0,0 +1,2 @@
+approvers:
+- stevekuznetsov

ci-operator/step-registry/ipi/conf/workload/ipi-conf-workload-commands.sh

@@ -0,0 +1,46 @@
+#!/bin/bash
+
+set -o nounset
+set -o errexit
+set -o pipefail
+
+# Run in the openshift namespace, to get access to the default ImageStreams.
+# https://docs.openshift.com/container-platform/4.6/openshift_images/using-imagestreams-with-kube-resources.html#images-managing-images-enabling-imagestreams-kube_using-imagestreams-with-kube-resources
+#cat >> "${SHARED_DIR}/manifest_workload-ns.yml" << EOF
+#apiVersion: v1
+#kind: Namespace
+#metadata:
+#  name: synthetic-workload
+#EOF
+
+cat >> "${SHARED_DIR}/manifest_loki-ss.yml" << EOF
+apiVersion: apps/v1
+kind: Deployment
+metadata:
+  name: synthetic-workload
+  namespace: openshift
+  annotations:
+    alpha.image.policy.openshift.io/resolve-names: '*'
+  selector:
+    matchLabels:
+      app: work
+spec:
+  replicas: 40  # this seems like it will be difficult to tune.  We want decent load on all of our update jobs, but no value will do that for 3-compute clusters while also working for compact, 0-compute clusters.
+  template:
+    metadata:
+      labels:
+        app: work
+    spec:
+      containers:
+      - name: work
+        image: openshift/cli:latest
+        command:
+        - "/bin/bash"
+        - "-c"
+        - 'i=1; while true; do if [[ "\$i" -gt 1 ]]; then i="\$((i-1))"; else i="\$((i+1))"; fi; done'
+        resources:
+          requests:  # FIXME: tune these to somewhat match reality
+            cpu: 100m
+            memory: 20Mi
+      terminationGracePeriodSeconds: 600
+EOF

ci-operator/step-registry/ipi/conf/workload/ipi-conf-workload-ref.metadata.json

@@ -0,0 +1,8 @@
+{
+	"path": "ipi/conf/workload/ipi-conf-workload-ref.yaml",
+	"owners": {
+		"approvers": [
+			"stevekuznetsov"
+		]
+	}
+}

ci-operator/step-registry/ipi/conf/workload/ipi-conf-workload-ref.yaml

@@ -0,0 +1,10 @@
+ref:
+  as: ipi-conf-workload
+  from: cli
+  commands: ipi-conf-workload-commands.sh
+  resources:
+    requests:
+      cpu: 100m
+      memory: 100Mi
+  documentation: |-
+    The workload configuration step deploys a synthetic workload to the cluster similar to the test-platform's slow-to-drain CI jobs.

ci-operator/step-registry/openshift/upgrade/aws/openshift-upgrade-aws-workflow.yaml

@@ -4,6 +4,7 @@ workflow:
     allow_best_effort_post_steps: true
     pre:
     - ref: ipi-install-loki
+    - ref: ipi-conf-workload
     - chain: ipi-aws-pre-stableinitial
     test:
     - ref: openshift-e2e-test