update canal with recent improvements to flannel (#77)

* align gh workflow and tox with current flannel

- dupe recent charm-flannel improvements for building resources
- drop crashdump in favor of a simple debug-log for failed runner jobs
- bump resource script to known-good charm-flannel commit
- tox tweaks to be consistent with other charms

* follow ck best practices for integration tests (overlay k8s-core with locally built charm/resources)

* use etcd3 calls to align with flannel changes

* dont run tests on l33t commits to main

.github/workflows/tox.yaml

@@ -1,52 +1,106 @@
 name: Run tests with Tox
 
 on:
-  push:
-    branches: [main]
   pull_request:
-    branches: [main]
+    branches:
+      - main
 
 jobs:
   call-inclusive-naming-check:
-    name: Inclusive Naming
+    name: Inclusive naming
     uses: canonical-web-and-design/Inclusive-naming/.github/workflows/woke.yaml@main
     with:
       fail-on-error: "true"
 
+  validate-wheelhouse:
+    name: Validate Wheelhouse
+    uses: charmed-kubernetes/workflows/.github/workflows/validate-wheelhouse.yaml@main
+
   lint-unit:
     name: Lint Unit
     uses: charmed-kubernetes/workflows/.github/workflows/lint-unit.yaml@main
 
-  validate-wheelhouse:
-    uses: charmed-kubernetes/workflows/.github/workflows/validate-wheelhouse.yaml@main  
+  resources-build:
+    name: Build Resources with docker
+    runs-on: ubuntu-22.04
+    needs:
+      - lint-unit
+    steps:
+      - name: Check out code
+        uses: actions/checkout@v3
+      - name: Install Docker
+        run:  sudo snap install docker
+      - name: Build Resources
+        run:  sudo ./build-canal-resources.sh
+      - name: Upload flannel artifact
+        uses: actions/upload-artifact@v3
+        with:
+          name: flannel-resources
+          path: ./flannel-*.tar.gz
+      - name: Upload calico artifact
+        uses: actions/upload-artifact@v3
+        with:
+          name: calico-resources
+          path: ./calico*.tar.gz
+      - name: Upload calico-node-image artifact
+        uses: actions/upload-artifact@v3
+        with:
+          name: calico-node-image-resources
+          path: ./calico-node-image.tar.gz
 
-  integration-test:
-    name: Integration test with VMWare
+  integration-tests:
+    name: Integration test with Vsphere
     runs-on: self-hosted
+    timeout-minutes: 60
     needs:
-      - call-inclusive-naming-check
-      - lint-unit
-      - validate-wheelhouse
-    timeout-minutes: 360
+      - resources-build
     steps:
       - name: Check out code
-        uses: actions/checkout@v2
+        uses: actions/checkout@v3
+
+      - name: Setup Python
+        uses: actions/setup-python@v4
+        with:
+          python-version: 3.8
+
       - name: Setup operator environment
         uses: charmed-kubernetes/actions-operator@main
         with:
           provider: vsphere
           credentials-yaml: ${{ secrets.CREDENTIALS_YAML }}
           clouds-yaml: ${{ secrets.CLOUDS_YAML }}
-          bootstrap-options: "--model-default datastore=vsanDatastore --model-default primary-network=VLAN_2764"
-      - name: Install docker (flannel resources)
-        run: sudo snap install docker
-      - name: Install juju-crashdump
-        run: sudo snap install juju-crashdump --classic
-      - name: Run test
+          bootstrap-options: "${{ secrets.FOCAL_BOOTSTRAP_OPTIONS }} --model-default datastore=vsanDatastore --model-default primary-network=VLAN_2764"
+          bootstrap-constraints: "arch=amd64 cores=2 mem=4G"
+
+      - name: Download flannel artifacts
+        uses: actions/download-artifact@v3
+        with:
+          name: flannel-resources
+
+      - name: Download calico artifacts
+        uses: actions/download-artifact@v3
+        with:
+          name: calico-resources
+
+      - name: Download calico-node-image artifacts
+        uses: actions/download-artifact@v3
+        with:
+          name: calico-node-image-resources
+
+      - name: Run integration test
         run: tox -e integration
-      - name: Upload crashdump
-        if: ${{ failure() }}
-        uses: actions/upload-artifact@v2
+
+      - name: Setup Debug Artifact Collection
+        if: failure()
+        run: mkdir tmp
+
+      - name: Collect Juju Logs
+        if: failure()
+        run: juju debug-log --replay --no-tail | tee tmp/juju-status.txt
+
+      - name: Upload debug artifacts
+        if: failure()
+        uses: actions/upload-artifact@v3
         with:
-          name: crashdump
-          path: juju-crashdump-*.tar.xz
+          name: test-run-artifacts
+          path: tmp

build-canal-resources.sh

@@ -21,14 +21,14 @@ mkdir -p "${canal_temp}"
 # FLANNEL RESOURCES:
 # The flannel version in the canal and flannel charms are the same; use flannel's
 # build-flannel-resource.sh so we build identical resources for the canal charm.
-FLANNEL_COMMIT="16d6841f019acc4ff6e5b04059ff4bee6b50057e"
+FLANNEL_COMMIT="db7a8123e2391c412462bb1cd5bce721ff8dd268"
 FLANNEL_REPO="https://github.com/charmed-kubernetes/charm-flannel.git"
 
 git clone $FLANNEL_REPO "${canal_temp}/flannel"
 pushd ${canal_temp}/flannel
 git checkout "$FLANNEL_COMMIT"
 ARCH="$ARCH" ./build-flannel-resources.sh
-mv flannel-*.gz ${canal_root}
+mv flannel-*.tar.gz ${canal_root}
 popd
 
 

reactive/flannel.py

@@ -171,13 +171,14 @@ def configure_network(etcd):
         }
     })
     cmd = "etcdctl "
-    cmd += "--endpoint '{0}' ".format(etcd.get_connection_string())
-    cmd += "--cert-file {0} ".format(ETCD_CERT_PATH)
-    cmd += "--key-file {0} ".format(ETCD_KEY_PATH)
-    cmd += "--ca-file {0} ".format(ETCD_CA_PATH)
-    cmd += "set /coreos.com/network/config '{0}'".format(data)
+    cmd += "--endpoints '{0}' ".format(etcd.get_connection_string())
+    cmd += "--cert {0} ".format(ETCD_CERT_PATH)
+    cmd += "--key {0} ".format(ETCD_KEY_PATH)
+    cmd += "--cacert {0} ".format(ETCD_CA_PATH)
+    cmd += "put /coreos.com/network/config '{0}'".format(data)
+    env = dict(os.environ, ETCDCTL_API="3")
     try:
-        check_call(split(cmd))
+        check_call(split(cmd), env=env)
         return True
 
     except CalledProcessError:

tests/data/charm.yaml

@@ -0,0 +1,26 @@
+description: A minimal Kubernetes cluster with two machines with virtual networks provided by Canal.
+series: &series {{ series }}
+applications:
+  calico: null
+  kubernetes-control-plane:
+    options:
+      channel: {{ snap_channel }}
+  kubernetes-worker:
+    options:
+      channel: {{ snap_channel }}
+  canal:
+    charm: {{ charm }}
+    channel: null
+    resources:
+      flannel: {{flannel_amd64|default("0")}}
+      flannel-arm64: {{flannel_arm64|default("0")}}
+      calico: {{calico_amd64|default("0")}}
+      calico-arm64: {{calico_arm64|default("0")}}
+      calico-node-image: {{calico_node_image|default("0")}}
+relations:
+- - canal:etcd
+  - etcd:db
+- - canal:cni
+  - kubernetes-control-plane:cni
+- - canal:cni
+  - kubernetes-worker:cni

tests/integration/conftest.py

@@ -1,56 +1,29 @@
 import logging
-from pathlib import Path
 import pytest
-import shlex
 
 log = logging.getLogger(__name__)
 
 
-def ensure_arch_names(current_resources):
-    for resource in current_resources:
-        if any(arch in resource.name for arch in ['s390x', 'arm64', 'amd64']):
-            continue
-        if resource.name == 'calico-node-image.tar.gz':
-            # doesn't have an associated arch
-            continue
-        # without an arch, assume 'amd64'
-        head, tail = resource.name.split('.', 1)
-        arched_name = "{}-amd64.{}".format(head, tail)
-        (resource.parent / arched_name).symlink_to(resource)
+def pytest_addoption(parser):
+    parser.addoption(
+        "--series",
+        type=str,
+        default="focal",
+        help="Set series for the machine units",
+    )
+    parser.addoption(
+        "--snap-channel",
+        type=str,
+        default="1.24/stable",
+        help="Set snap channel for the control-plane & worker units",
+    )
 
 
 @pytest.fixture()
-async def setup_resources(ops_test):
-    """Provides the flannel resources needed to deploy the charm."""
-    script_path = resource_path = Path.cwd()
-    current_resources = list(resource_path.glob("*.tar.gz"))
-    tmpdir = ops_test.tmp_path / "resources"
-    tmpdir.mkdir(parents=True, exist_ok=True)
-    if not current_resources:
-        # If they are not locally available, try to build them
-        log.info("Build Resources...")
-        build_script = script_path / "build-canal-resources.sh"
-        rc, stdout, stderr = await ops_test.run(
-            *shlex.split("sudo {}".format(build_script)), cwd=tmpdir, check=False
-        )
-        if rc != 0:
-            err = (stderr or stdout).strip()
-            log.warning("build-flannel-resources failed: {}".format(err))
-        current_resources = list(Path(tmpdir).glob("*.tar.gz"))
-        resource_path = tmpdir
-    if not current_resources:
-        # if we couldn't build them, just download a fixed version
-        log.info("Downloading Resources...")
-        fetch_script = script_path / "fetch-charm-store-resources.sh"
-        rc, stdout, stderr = await ops_test.run(
-            *shlex.split(str(fetch_script)), cwd=tmpdir, check=False
-        )
-        if rc != 0:
-            err = (stderr or stdout).strip()
-            log.warning("fetch-charm-store-resources failed: {}".format(err))
-        current_resources = list(Path(tmpdir).glob("*.tar.gz"))
-        resource_path = tmpdir
-    if not current_resources:
-        pytest.fail("Could not prepare necessary resources for testing charm")
-    ensure_arch_names(current_resources)
-    yield resource_path
+def series(request):
+    return request.config.getoption("--series")
+
+
+@pytest.fixture()
+def snap_channel(request):
+    return request.config.getoption("--snap-channel")