RHOAIENG-16076: tests(gha): install a cri-o backed kubernetes for running Makefile tests (#783)

jiridanek · jstourac · web-flow · commit 117a47931673 · 2024-11-28T05:23:19.000Z
* RHOAIENG-16076: tests(gha): install a cri-o backed kubernetes for running tests

* Update .github/workflows/build-notebooks-TEMPLATE.yaml

Co-authored-by: Jan Stourac &lt;jstourac@redhat.com&gt;

---------

Co-authored-by: Jan Stourac &lt;jstourac@redhat.com&gt;
diff --git a/.github/workflows/build-notebooks-TEMPLATE.yaml b/.github/workflows/build-notebooks-TEMPLATE.yaml
@@ -235,6 +235,121 @@ jobs:
       - name: "Show podman images information"
         run: podman images --digests
 
+      - name: "Check if we have tests or not"
+        id: have-tests
+        run: "ci/cached-builds/has_tests.py --target ${{ inputs.target }}"
+
+      # https://cri-o.io/
+      - name: Install cri-o
+        if: ${{ steps.have-tests.outputs.tests == 'true' }}
+        run: |
+          set -Eeuxo pipefail
+
+          sudo apt-get update
+          sudo apt-get install -y software-properties-common curl
+
+          curl -fsSL https://pkgs.k8s.io/core:/stable:/$KUBERNETES_VERSION/deb/Release.key | \
+            sudo gpg --dearmor --batch --yes -o /etc/apt/keyrings/kubernetes-apt-keyring.gpg
+
+          echo "deb [signed-by=/etc/apt/keyrings/kubernetes-apt-keyring.gpg] https://pkgs.k8s.io/core:/stable:/$KUBERNETES_VERSION/deb/ /" | \
+            sudo tee /etc/apt/sources.list.d/kubernetes.list
+
+          curl -fsSL https://pkgs.k8s.io/addons:/cri-o:/stable:/$CRIO_VERSION/deb/Release.key | \
+            sudo gpg --dearmor --batch --yes -o /etc/apt/keyrings/cri-o-apt-keyring.gpg
+
+          echo "deb [signed-by=/etc/apt/keyrings/cri-o-apt-keyring.gpg] https://pkgs.k8s.io/addons:/cri-o:/stable:/$CRIO_VERSION/deb/ /" | \
+            sudo tee /etc/apt/sources.list.d/cri-o.list
+
+          sudo apt-get update
+          sudo apt-get install -y cri-o kubelet kubeadm kubectl
+
+          # make use of /etc/cni/net.d/11-crio-ipv4-bridge.conflist so we don't
+          # need a pod network and just use the default bridge
+          sudo rm -rf /etc/cni/net.d/*
+          # cat /etc/cni/net.d/11-crio-ipv4-bridge.conflist
+          # https://github.com/containerd/containerd/blob/main/script%2Fsetup%2Finstall-cni
+          # https://www.cni.dev/plugins/current/main/bridge/
+          sudo cp ci/cached-builds/11-crio-ipv4-bridge.conflist /etc/cni/net.d/11-crio-ipv4-bridge.conflist
+
+          sudo cp ci/cached-builds/crio.conf /etc/crio/crio.conf.d/
+
+          sudo systemctl start crio.service
+        env:
+          CRIO_VERSION: v1.30
+          KUBERNETES_VERSION: v1.30
+
+      - name: Show crio debug data (on failure)
+        if: ${{ failure() && steps.have-tests.outputs.tests == 'true' }}
+        run: |
+          set -Eeuxo pipefail
+
+          sudo systemctl status crio.service || true
+          sudo journalctl -xeu crio.service
+
+      # do this early, it's a good check that cri-o is not completely broken
+      - name: "Show crio images information"
+        if: ${{ steps.have-tests.outputs.tests == 'true'  }}
+        run: sudo crictl images
+
+      - name: Install Kubernetes cluster
+        if: ${{ steps.have-tests.outputs.tests == 'true'  }}
+        run: |
+          set -Eeuxo pipefail
+
+          sudo swapoff -a
+          sudo modprobe br_netfilter
+          sudo sysctl -w net.ipv4.ip_forward=1
+
+          # Was getting strange DNS resolution errors from pods that don't seem to want to go away sometimes:
+          #  Resolving raw.githubusercontent.com (raw.githubusercontent.com)... failed: Name or service not known.
+          #  wget: unable to resolve host address ‘raw.githubusercontent.com’
+          # Here's what helped:
+          #  https://kubernetes.io/docs/tasks/administer-cluster/dns-debugging-resolution/#known-issues
+          #  https://github.com/kubernetes/kubernetes/blob/e4c1f980b76fecece30c2f77885a7117192170a6/CHANGELOG/CHANGELOG-1.30.md?plain=1#L1454
+          #  https://github.com/canonical/microk8s/issues/68#issuecomment-404923563
+          sudo ufw allow in on cni0
+          sudo ufw allow out on cni0
+          sudo ufw default allow routed
+          sudo iptables -P FORWARD ACCEPT
+          sudo iptables -t nat -A POSTROUTING -s 10.85.0.0/16 -o eth0 -j MASQUERADE
+
+          # https://kubernetes.io/docs/setup/production-environment/tools/kubeadm/create-cluster-kubeadm
+          sudo kubeadm init --config=ci/cached-builds/kubeadm.yaml
+
+          mkdir -p $HOME/.kube
+          sudo cp -i /etc/kubernetes/admin.conf $HOME/.kube/config
+          sudo chown $(id -u):$(id -g) $HOME/.kube/config
+
+      - name: Show kubelet debug data (on failure)
+        if: ${{ failure() && steps.have-tests.outputs.tests == 'true' }}
+        run: |
+          set -Eeuxo pipefail
+
+          sudo systemctl status kubelet || true
+          sudo journalctl -xeu kubelet
+
+          # Here is one example how you may list all running Kubernetes containers by using crictl:
+          sudo crictl --runtime-endpoint unix:///var/run/crio/crio.sock ps -a | grep kube | grep -v pause
+          # Once you have found the failing container, you can inspect its logs with:
+          # crictl --runtime-endpoint unix:///var/run/crio/crio.sock logs CONTAINERID
+
+      - name: Untaint the master
+        if: ${{ steps.have-tests.outputs.tests == 'true' }}
+        run: kubectl taint nodes --all node-role.kubernetes.io/control-plane-
+
+      - name: Show nodes status and wait for readiness
+        if: ${{ steps.have-tests.outputs.tests == 'true' }}
+        run: |
+          kubectl describe nodes
+          kubectl wait --for=condition=Ready nodes --all --timeout=100s || (kubectl describe nodes && false)
+
+      - name: Wait for pods to be running
+        if: ${{ steps.have-tests.outputs.tests == 'true' }}
+        run: |
+          set -Eeuxo pipefail
+          kubectl wait deployments --all --all-namespaces --for=condition=Available --timeout=100s
+          kubectl wait pods --all --all-namespaces --for=condition=Ready --timeout=100s
+
       - name: Run Trivy vulnerability scanner
         if: ${{ steps.resolve-target.outputs.target }}
         run: |
diff --git a/ci/cached-builds/11-crio-ipv4-bridge.conflist b/ci/cached-builds/11-crio-ipv4-bridge.conflist
@@ -0,0 +1,22 @@
+{
+  "cniVersion": "1.0.0",
+  "name": "crio",
+  "plugins": [
+    {
+      "type": "bridge",
+      "bridge": "cni0",
+      "isGateway": true,
+      "ipMasq": true,
+      "hairpinMode": true,
+      "ipam": {
+        "type": "host-local",
+        "routes": [
+            { "dst": "0.0.0.0/0" }
+        ],
+        "ranges": [
+            [{ "subnet": "10.85.0.0/16" }]
+        ]
+      }
+    }
+  ]
+}
diff --git a/ci/cached-builds/crio.conf b/ci/cached-builds/crio.conf
@@ -0,0 +1,17 @@
+# https://github.com/cri-o/cri-o/blob/main/docs/crio.conf.5.md
+
+[crio]
+storage_driver = "overlay"
+# storage_option = [ "overlay.mountopt=nodev,metacopy=on" ]
+
+# reuse podman's container storage because we have huge images that don't fit on disk twice
+root = "/home/runner/.local/share/containers/storage"
+# has to be the same as root!
+runroot = "/home/runner/.local/share/containers/storage"
+
+# https://stackoverflow.com/questions/62408028/kubelet-failed-to-createpodsandbox-for-coredns-failed-to-set-bridge-addr-c
+[crio.network]
+# the /etc/cni/net.d/11-crio-ipv4-bridge.conflist default IPs confilct with flannel,
+# older versions of kubernetes the kubelet was touching the cni, now only the container runtime touches
+# c.f. https://kubernetes.io/docs/concepts/extend-kubernetes/compute-storage-net/network-plugins/#installation
+#network_dir = "/etc/cni/net.d-kube/"
diff --git a/ci/cached-builds/has_tests.py b/ci/cached-builds/has_tests.py
@@ -0,0 +1,58 @@
+#!/usr/bin/env python3
+import argparse
+import json
+import os
+import pathlib
+import typing
+import unittest
+
+import gha_pr_changed_files
+
+"""Determines whether we have deploy Makefile tests for this target or not
+
+https://github.com/openshift/release/blob/master/ci-operator/config/opendatahub-io/notebooks/opendatahub-io-notebooks-main.yaml#L1485
+"""
+
+
+class Args(argparse.Namespace):
+    """Type annotation to have autocompletion for args"""
+    target: str
+
+
+def main() -> None:
+    parser = argparse.ArgumentParser("make_test.py")
+    parser.add_argument("--target", type=str)
+    args = typing.cast(Args, parser.parse_args())
+
+    has_tests = check_tests(args.target)
+
+    if "GITHUB_ACTIONS" in os.environ:
+        with open(os.environ["GITHUB_OUTPUT"], "at") as f:
+            print(f"tests={json.dumps(has_tests)}", file=f)
+
+    print(f"{has_tests=}")
+
+
+def check_tests(target: str) -> bool:
+    if target.startswith("rocm-jupyter-minimal-") or target.startswith("rocm-jupyter-datascience-"):
+        return False  # we don't have specific tests for -minimal-, ... in ci-operator/config
+    if '-intel-' in target:
+        return False  # RHOAIENG-8388: Intel tensorflow notebook failed to get tested on OCP-CI
+
+    has_tests = False
+    dirs = gha_pr_changed_files.analyze_build_directories(target)
+    for d in reversed(dirs):  # (!)
+        kustomization = pathlib.Path(gha_pr_changed_files.PROJECT_ROOT) / d / "kustomize/base/kustomization.yaml"
+        has_tests = has_tests or kustomization.is_file()
+        break  # TODO: check only the last directory (the top level layer) for now
+    return has_tests
+
+
+class TestCheckTests(unittest.TestCase):
+    def test_has_tests(self):
+        assert check_tests("base-c9s-python-3.11") is False
+        assert check_tests("jupyter-minimal-ubi9-python-3.9") is True
+
+
+if __name__ == "__main__":
+    main()
diff --git a/ci/cached-builds/kubeadm.yaml b/ci/cached-builds/kubeadm.yaml
@@ -0,0 +1,45 @@
+---
+# kubeadm config print init-defaults > kubeadm.yaml
+# kubeadm init --cri-socket=/var/run/crio/crio.sock
+
+# https://kubernetes.io/docs/reference/config-api/kubeadm-config.v1beta3/
+# https://kubernetes.io/docs/reference/config-api/kubeadm-config.v1beta4/
+apiVersion: kubeadm.k8s.io/v1beta3
+bootstrapTokens:
+  - groups:
+      - system:bootstrappers:kubeadm:default-node-token
+    token: abcdef.0123456789abcdef
+    ttl: 24h0m0s
+    usages:
+      - signing
+      - authentication
+kind: InitConfiguration
+localAPIEndpoint:
+  bindPort: 6443
+nodeRegistration:
+  kubeletExtraArgs:
+    # Need to have enough disk space for Kubelet, so move root-dir on the LVM volume
+    # Note: the internets discourage from changing the default because storage plugins may then struggle
+    #  https://cep.dev/posts/adventure-trying-change-kubelet-rootdir/
+    root-dir: "/home/runner/.local/share/containers/kubelet-root-dir"
+  criSocket: unix:///var/run/crio/crio.sock
+  imagePullPolicy: IfNotPresent
+  taints: null
+---
+apiServer:
+  timeoutForControlPlane: 4m0s
+apiVersion: kubeadm.k8s.io/v1beta3
+certificatesDir: /etc/kubernetes/pki
+clusterName: kubernetes
+controllerManager: {}
+dns: {}
+etcd:
+  local:
+    dataDir: /var/lib/etcd
+imageRepository: registry.k8s.io
+kind: ClusterConfiguration
+networking:
+  dnsDomain: cluster.local
+  # this matches the default in /etc/cni/net.d/11-crio-ipv4-bridge.conflist
+  podSubnet: 10.85.0.0/16
+scheduler: {}
