Run integration tests on arm64 (#450)

Author: carlcsaposs-canonical

.github/workflows/ci.yaml

@@ -65,24 +65,32 @@ jobs:
       fail-fast: false
       matrix:
         juju:
-          - agent: 2.9.49
+          - agent: 2.9.49  # renovate: juju-agent-pin-minor
             libjuju: ^2
-            allure: false
-          - agent: 3.4.3
-            allure: true
-    name: Integration test charm | ${{ matrix.juju.agent }}
+            allure_on_amd64: false
+          - agent: 3.4.3  # renovate: juju-agent-pin-minor
+            allure_on_amd64: true
+        architecture:
+          - amd64
+        include:
+          - juju:
+              agent: 3.4.3  # renovate: juju-agent-pin-minor
+              allure_on_amd64: true
+            architecture: arm64
+    name: Integration test charm | ${{ matrix.juju.agent }} | ${{ matrix.architecture }}
     needs:
       - lint
       - unit-test
       - build
     uses: canonical/data-platform-workflows/.github/workflows/[email protected]
     with:
       artifact-prefix: ${{ needs.build.outputs.artifact-prefix }}
+      architecture: ${{ matrix.architecture }}
       cloud: microk8s
       microk8s-snap-channel: 1.28-strict/stable
       juju-agent-version: ${{ matrix.juju.agent }}
       libjuju-version-constraint: ${{ matrix.juju.libjuju }}
-      _beta_allure_report: ${{ matrix.juju.allure }}
+      _beta_allure_report: ${{ matrix.juju.allure_on_amd64 && matrix.architecture == 'amd64' }}
     secrets:
       # GitHub appears to redact each line of a multi-line secret
       # Avoid putting `{` or `}` on a line by itself so that it doesn't get redacted in logs

tests/integration/architecture.py

@@ -0,0 +1,7 @@
+# Copyright 2024 Canonical Ltd.
+# See LICENSE file for licensing details.
+import subprocess
+
+architecture = subprocess.run(
+    ["dpkg", "--print-architecture"], capture_output=True, check=True, encoding="utf-8"
+).stdout.strip()

tests/integration/helpers.py

@@ -3,6 +3,7 @@
 
 import itertools
 import json
+import pathlib
 import secrets
 import string
 import subprocess
@@ -556,13 +557,13 @@ async def write_content_to_file_in_unit(
     """
     pod_name = unit.name.replace("/", "-")
 
-    with tempfile.NamedTemporaryFile(mode="w") as temp_file:
+    with tempfile.NamedTemporaryFile(mode="w", dir=pathlib.Path.home()) as temp_file:
         temp_file.write(content)
         temp_file.flush()
 
         subprocess.run(
             [
-                "kubectl",
+                "microk8s.kubectl",
                 "cp",
                 "-n",
                 ops_test.model.info.name,
@@ -591,10 +592,10 @@ async def read_contents_from_file_in_unit(
     """
     pod_name = unit.name.replace("/", "-")
 
-    with tempfile.NamedTemporaryFile(mode="r+") as temp_file:
+    with tempfile.NamedTemporaryFile(mode="r+", dir=pathlib.Path.home()) as temp_file:
         subprocess.run(
             [
-                "kubectl",
+                "microk8s.kubectl",
                 "cp",
                 "-n",
                 ops_test.model.info.name,

tests/integration/high_availability/high_availability_helpers.py

@@ -259,7 +259,7 @@ def deploy_chaos_mesh(namespace: str) -> None:
         for attempt in Retrying(stop=stop_after_delay(5 * 60), wait=wait_fixed(10)):
             with attempt:
                 output = subprocess.check_output(
-                    f"kubectl get pods --namespace {namespace} -l app.kubernetes.io/instance=chaos-mesh".split(),
+                    f"microk8s.kubectl get pods --namespace {namespace} -l app.kubernetes.io/instance=chaos-mesh".split(),
                     env=env,
                 )
                 assert output.decode().count("Running") == 4, "Chaos Mesh not ready"
@@ -528,7 +528,7 @@ def isolate_instance_from_cluster(ops_test: OpsTest, unit_name: str) -> None:
         env["KUBECONFIG"] = os.path.expanduser("~/.kube/config")
 
         try:
-            subprocess.check_output(["kubectl", "apply", "-f", temp_file.name], env=env)
+            subprocess.check_output(["microk8s.kubectl", "apply", "-f", temp_file.name], env=env)
         except subprocess.CalledProcessError as e:
             logger.error(e.output)
             logger.error(e.stderr)
@@ -540,7 +540,7 @@ def remove_instance_isolation(ops_test: OpsTest) -> None:
     env = os.environ
     env["KUBECONFIG"] = os.path.expanduser("~/.kube/config")
     subprocess.check_output(
-        f"kubectl -n {ops_test.model.info.name} delete networkchaos network-loss-primary",
+        f"microk8s.kubectl -n {ops_test.model.info.name} delete networkchaos network-loss-primary",
         shell=True,
         env=env,
     )

tests/integration/high_availability/scripts/destroy_chaos_mesh.sh

@@ -11,38 +11,38 @@ fi
 
 destroy_chaos_mesh() {
 	echo "deleting api-resources"
-	for i in $(kubectl api-resources | awk '/chaos-mesh/ {print $1}'); do
-	    timeout 30 kubectl delete "${i}" --all --all-namespaces || true
+	for i in $(microk8s.kubectl api-resources | awk '/chaos-mesh/ {print $1}'); do
+	    timeout 30 microk8s.kubectl delete "${i}" --all --all-namespaces || true
 	done
 
-	if kubectl get mutatingwebhookconfiguration | grep -q 'chaos-mesh-mutation'; then
-		timeout 30 kubectl delete mutatingwebhookconfiguration chaos-mesh-mutation || true
+	if microk8s.kubectl get mutatingwebhookconfiguration | grep -q 'chaos-mesh-mutation'; then
+		timeout 30 microk8s.kubectl delete mutatingwebhookconfiguration chaos-mesh-mutation || true
 	fi
 
-	if kubectl get validatingwebhookconfiguration | grep -q 'chaos-mesh-validation-auth'; then
-		timeout 30 kubectl delete validatingwebhookconfiguration chaos-mesh-validation-auth || true
+	if microk8s.kubectl get validatingwebhookconfiguration | grep -q 'chaos-mesh-validation-auth'; then
+		timeout 30 microk8s.kubectl delete validatingwebhookconfiguration chaos-mesh-validation-auth || true
 	fi
 
-	if kubectl get validatingwebhookconfiguration | grep -q 'chaos-mesh-validation'; then
-		timeout 30 kubectl delete validatingwebhookconfiguration chaos-mesh-validation || true
+	if microk8s.kubectl get validatingwebhookconfiguration | grep -q 'chaos-mesh-validation'; then
+		timeout 30 microk8s.kubectl delete validatingwebhookconfiguration chaos-mesh-validation || true
 	fi
 
-	if kubectl get clusterrolebinding | grep -q 'chaos-mesh'; then
+	if microk8s.kubectl get clusterrolebinding | grep -q 'chaos-mesh'; then
 		echo "deleting clusterrolebindings"
-		readarray -t args < <(kubectl get clusterrolebinding | awk '/chaos-mesh/ {print $1}')
-		timeout 30 kubectl delete clusterrolebinding "${args[@]}" || true
+		readarray -t args < <(microk8s.kubectl get clusterrolebinding | awk '/chaos-mesh/ {print $1}')
+		timeout 30 microk8s.kubectl delete clusterrolebinding "${args[@]}" || true
 	fi
 
-	if kubectl get clusterrole | grep -q 'chaos-mesh'; then
+	if microk8s.kubectl get clusterrole | grep -q 'chaos-mesh'; then
 		echo "deleting clusterroles"
-		readarray -t args < <(kubectl get clusterrole | awk '/chaos-mesh/ {print $1}')
-		timeout 30 kubectl delete clusterrole "${args[@]}" || true
+		readarray -t args < <(microk8s.kubectl get clusterrole | awk '/chaos-mesh/ {print $1}')
+		timeout 30 microk8s.kubectl delete clusterrole "${args[@]}" || true
 	fi
 
-	if kubectl get crd | grep -q 'chaos-mesh.org'; then
+	if microk8s.kubectl get crd | grep -q 'chaos-mesh.org'; then
 		echo "deleting crds"
-		readarray -t args < <(kubectl get crd | awk '/chaos-mesh.org/ {print $1}')
-		timeout 30 kubectl delete crd "${args[@]}" || true
+		readarray -t args < <(microk8s.kubectl get crd | awk '/chaos-mesh.org/ {print $1}')
+		timeout 30 microk8s.kubectl delete crd "${args[@]}" || true
 	fi
 
 	if [ -n "${chaos_mesh_ns}" ] && sg snap_microk8s -c "microk8s.helm3 repo list --namespace=${chaos_mesh_ns}" | grep -q 'chaos-mesh'; then

tests/integration/high_availability/test_async_replication.py

@@ -4,6 +4,7 @@
 
 
 import logging
+import subprocess
 from asyncio import gather
 from pathlib import Path
 from time import sleep
@@ -14,14 +15,13 @@
 from juju.model import Model
 from pytest_operator.plugin import OpsTest
 
-from .. import juju_
+from .. import architecture, juju_, markers
 from ..helpers import (
     execute_queries_on_unit,
     get_cluster_status,
     get_leader_unit,
     get_unit_address,
 )
-from ..markers import juju3
 from .high_availability_helpers import (
     DATABASE_NAME,
     TABLE_NAME,
@@ -55,6 +55,11 @@ async def second_model(
     """Create and return the second model."""
     second_model_name = f"{first_model.info.name}-other"
     await ops_test._controller.add_model(second_model_name)
+    subprocess.run(["juju", "switch", second_model_name], check=True)
+    subprocess.run(
+        ["juju", "set-model-constraints", f"arch={architecture.architecture}"], check=True
+    )
+    subprocess.run(["juju", "switch", first_model.info.name], check=True)
     second_model = Model()
     await second_model.connect(model_name=second_model_name)
     yield second_model  # pyright: ignore [reportReturnType]
@@ -65,9 +70,10 @@ async def second_model(
     await ops_test._controller.destroy_model(second_model_name, destroy_storage=True)
 
 
-@juju3
-@pytest.mark.abort_on_fail
 @pytest.mark.group(1)
+@markers.juju3
+@markers.amd64_only  # TODO: remove after mysql-router-k8s arm64 stable release
+@pytest.mark.abort_on_fail
 async def test_build_and_deploy(
     ops_test: OpsTest, first_model: Model, second_model: Model
 ) -> None:
@@ -112,9 +118,10 @@ async def test_build_and_deploy(
     )
 
 
-@juju3
-@pytest.mark.abort_on_fail
 @pytest.mark.group(1)
+@markers.juju3
+@markers.amd64_only  # TODO: remove after mysql-router-k8s arm64 stable release
+@pytest.mark.abort_on_fail
 async def test_async_relate(first_model: Model, second_model: Model) -> None:
     """Relate the two mysql clusters."""
     logger.info("Creating offers in first model")
@@ -145,9 +152,10 @@ async def test_async_relate(first_model: Model, second_model: Model) -> None:
     )
 
 
-@juju3
-@pytest.mark.abort_on_fail
 @pytest.mark.group(1)
+@markers.juju3
+@markers.amd64_only  # TODO: remove after mysql-router-k8s arm64 stable release
+@pytest.mark.abort_on_fail
 async def test_create_replication(first_model: Model, second_model: Model) -> None:
     """Run the create replication and wait for the applications to settle."""
     logger.info("Running create replication action")
@@ -175,9 +183,10 @@ async def test_create_replication(first_model: Model, second_model: Model) -> No
     )
 
 
-@juju3
-@pytest.mark.abort_on_fail
 @pytest.mark.group(1)
+@markers.juju3
+@markers.amd64_only  # TODO: remove after mysql-router-k8s arm64 stable release
+@pytest.mark.abort_on_fail
 async def test_deploy_router_and_app(first_model: Model) -> None:
     """Deploy the router and the test application."""
     logger.info("Deploying router and application")
@@ -193,7 +202,7 @@ async def test_deploy_router_and_app(first_model: Model) -> None:
         APPLICATION_APP_NAME,
         application_name=APPLICATION_APP_NAME,
         series="jammy",
-        channel="latest/stable",
+        channel="latest/edge",
         num_units=1,
     )
 
@@ -212,9 +221,10 @@ async def test_deploy_router_and_app(first_model: Model) -> None:
     )
 
 
-@juju3
-@pytest.mark.abort_on_fail
 @pytest.mark.group(1)
+@markers.juju3
+@markers.amd64_only  # TODO: remove after mysql-router-k8s arm64 stable release
+@pytest.mark.abort_on_fail
 async def test_data_replication(
     first_model: Model, second_model: Model, continuous_writes
 ) -> None:
@@ -225,9 +235,10 @@ async def test_data_replication(
     assert results[0] > 1, "No data was written to the database"
 
 
-@juju3
-@pytest.mark.abort_on_fail
 @pytest.mark.group(1)
+@markers.juju3
+@markers.amd64_only  # TODO: remove after mysql-router-k8s arm64 stable release
+@pytest.mark.abort_on_fail
 async def test_standby_promotion(
     ops_test: OpsTest, first_model: Model, second_model: Model, continuous_writes
 ) -> None:
@@ -253,9 +264,10 @@ async def test_standby_promotion(
     ), "standby not promoted to primary"
 
 
-@juju3
-@pytest.mark.abort_on_fail
 @pytest.mark.group(1)
+@markers.juju3
+@markers.amd64_only  # TODO: remove after mysql-router-k8s arm64 stable release
+@pytest.mark.abort_on_fail
 async def test_failover(ops_test: OpsTest, first_model: Model, second_model: Model) -> None:
     """Test switchover on primary cluster fail."""
     logger.info("Freezing mysqld on primary cluster units")
@@ -291,9 +303,10 @@ async def test_failover(ops_test: OpsTest, first_model: Model, second_model: Mod
         )
 
 
-@juju3
-@pytest.mark.abort_on_fail
 @pytest.mark.group(1)
+@markers.juju3
+@markers.amd64_only  # TODO: remove after mysql-router-k8s arm64 stable release
+@pytest.mark.abort_on_fail
 async def test_rejoin_invalidated_cluster(
     first_model: Model, second_model: Model, continuous_writes
 ) -> None:
@@ -311,9 +324,10 @@ async def test_rejoin_invalidated_cluster(
     assert results[0] > 1, "No data was written to the database"
 
 
-@juju3
-@pytest.mark.abort_on_fail
 @pytest.mark.group(1)
+@markers.juju3
+@markers.amd64_only  # TODO: remove after mysql-router-k8s arm64 stable release
+@pytest.mark.abort_on_fail
 async def test_remove_relation_and_relate(
     first_model: Model, second_model: Model, continuous_writes
 ) -> None:

tests/integration/high_availability/test_upgrade_from_stable.py

@@ -8,7 +8,7 @@
 import pytest
 from pytest_operator.plugin import OpsTest
 
-from .. import juju_
+from .. import juju_, markers
 from ..helpers import (
     get_leader_unit,
     get_primary_unit,
@@ -31,6 +31,7 @@
 
 
 @pytest.mark.group(1)
+@markers.amd64_only  # TODO: remove after arm64 stable release
 @pytest.mark.abort_on_fail
 async def test_deploy_stable(ops_test: OpsTest) -> None:
     """Simple test to ensure that the mysql and application charms get deployed."""
@@ -61,6 +62,7 @@ async def test_deploy_stable(ops_test: OpsTest) -> None:
 
 
 @pytest.mark.group(1)
+@markers.amd64_only  # TODO: remove after arm64 stable release
 @pytest.mark.abort_on_fail
 async def test_pre_upgrade_check(ops_test: OpsTest) -> None:
     """Test that the pre-upgrade-check action runs successfully."""
@@ -87,6 +89,7 @@ async def test_pre_upgrade_check(ops_test: OpsTest) -> None:
 
 
 @pytest.mark.group(1)
+@markers.amd64_only  # TODO: remove after arm64 stable release
 @pytest.mark.abort_on_fail
 async def test_upgrade_from_stable(ops_test: OpsTest):
     """Test updating from stable channel."""