CSS-7092: Add server upgrade test

.github/workflows/integration_test.yaml

@@ -15,6 +15,7 @@ jobs:
           - integration-scaling
           - integration-upgrades
           - integration-auth
+          - integration-server-upgrade
     runs-on: ubuntu-latest
     steps:
       - name: Checkout

documentation/how-to/upgrade-server.md

@@ -33,8 +33,8 @@ The Temporal K8s charms facilitate server upgrades in the following way:
    This will ensure that your database schema is updated if any updates are
    available.
 
-2. The Temporal K8s charm should be updated to the next charm revision
-   that you currently have deployed as follows:
+2. The Temporal K8s charm should be updated to the next charm revision that you
+   currently have deployed as follows:
 
    ```bash
    juju refresh temporal-k8s --revision=<your_revision + 1>
@@ -44,3 +44,14 @@ _Warning: It is essential that upgrades are done one consecutive revision at a
 time. Charmed Temporal K8s can only guarantee backward compatibility between two
 consecutive revisions in line with the upgrade system adopted by the Temporal
 Server._
+
+## Appendix
+
+The table below shows a mapping between the Temporal K8s charms and the Temporal
+server versions. It can be used as a reference for upgrading your charm
+revisions in line with the Temporal server version so as to avoid any breaking
+changes.
+
+| Temporal Server Charm Revision | Temporal Admin Charm Revision | Temporal Server Version |
+| :----------------------------: | :---------------------------: | :---------------------: |
+|             20-21              |              8-9              |         v1.21.5         |

templates/config.jinja

@@ -291,5 +291,5 @@ additionalEnvs:
 {%- endif %}
 
 dynamicConfigClient:
-    filepath: "{{ DYNAMIC_CONFIG_FILE_PATH | default("/etc/temporal/config/dynamicconfig/development.yaml") }}"
+    filepath: "{{ DYNAMIC_CONFIG_FILE_PATH | default("/etc/temporal/config/dynamicconfig/docker.yaml") }}"
     pollInterval: "60s"

tests/integration/test_server_upgrade.py

@@ -0,0 +1,103 @@
+# Copyright 2024 Canonical Ltd.
+# See LICENSE file for licensing details.
+
+"""Temporal charm upgrades integration tests."""
+
+import asyncio
+import logging
+import time
+
+import pytest
+import pytest_asyncio
+from helpers import (
+    APP_NAME,
+    APP_NAME_ADMIN,
+    APP_NAME_UI,
+    create_default_namespace,
+    perform_temporal_integrations,
+    run_sample_workflow,
+)
+from pytest_operator.plugin import OpsTest
+
+logger = logging.getLogger(__name__)
+
+
+@pytest.mark.skip_if_deployed
+@pytest_asyncio.fixture(name="deploy", scope="module")
+async def deploy(ops_test: OpsTest):
+    """The app is up and running."""
+    await ops_test.model.set_config({"update-status-hook-interval": "1m"})
+    charm = await ops_test.build_charm(".")
+
+    # Deploy Temporal server, Temporal admin, Temporal UI and postgresql charms.
+    asyncio.gather(
+        ops_test.model.deploy(
+            charm,
+            application_name=APP_NAME,
+            resources={"temporal-server-image": "temporalio/server:1.20.0"},
+            config={"num-history-shards": "1"},
+        ),
+        ops_test.model.deploy(
+            APP_NAME_ADMIN, channel="edge", resources={"temporal-admin-image": "temporalio/admin-tools:1.20.0"}
+        ),
+        ops_test.model.deploy(APP_NAME_UI, channel="edge"),
+        ops_test.model.deploy("postgresql-k8s", channel="14/stable", trust=True),
+    )
+
+    async with ops_test.fast_forward():
+        await ops_test.model.wait_for_idle(
+            apps=[APP_NAME, APP_NAME_ADMIN, APP_NAME_UI], status="blocked", raise_on_blocked=False, timeout=600
+        )
+        await ops_test.model.wait_for_idle(
+            apps=["postgresql-k8s"], status="active", raise_on_blocked=False, timeout=600
+        )
+
+        await perform_temporal_integrations(ops_test)
+
+        await create_default_namespace(ops_test)
+
+        await ops_test.model.wait_for_idle(apps=[APP_NAME], status="active", raise_on_blocked=False, timeout=300)
+        assert ops_test.model.applications[APP_NAME].units[0].workload_status == "active"
+        assert ops_test.model.applications[APP_NAME_UI].units[0].workload_status == "active"
+        await run_sample_workflow(ops_test)
+
+
+@pytest.mark.abort_on_fail
+@pytest.mark.usefixtures("deploy")
+class TestServerUpgrade:
+    """Integration test for Temporal server upgrade requiring schema update.
+
+    This test ensures that upgrading from v1.20.0 to v1.21.2 (which requires a schema update) runs
+    successfully on the newly built charm.
+    """
+
+    async def test_server_upgrade(self, ops_test: OpsTest):
+        """Refresh the charm with a new resource which requires a schema update."""
+        # Update admin charm to v1.21.2 first
+        await ops_test.model.applications[APP_NAME_ADMIN].refresh(
+            resources={"temporal-admin-image": "temporalio/admin-tools:1.21.2"},
+        )
+        await ops_test.model.wait_for_idle(
+            apps=[APP_NAME_ADMIN], raise_on_error=False, status="active", raise_on_blocked=False, timeout=600
+        )
+
+        # Needed for a local charm refresh
+        charm = await ops_test.build_charm(".")
+
+        # Update server charm to v1.21.2
+        await ops_test.model.applications[APP_NAME].refresh(
+            resources={"temporal-server-image": "temporalio/server:1.21.2"},
+            path=str(charm),
+        )
+
+        # This is to accmmodate for a self-resolving error which sometimes appears when Temporal
+        # services attempt to connect to the cluster before the application is ready.
+        await ops_test.model.wait_for_idle(
+            apps=[APP_NAME], raise_on_error=False, status="active", raise_on_blocked=False, timeout=600
+        )
+        time.sleep(10)
+
+        assert ops_test.model.applications[APP_NAME].units[0].workload_status == "active"
+        assert ops_test.model.applications[APP_NAME_ADMIN].units[0].workload_status == "active"
+
+        await run_sample_workflow(ops_test)

tests/integration/test_upgrades.py

@@ -29,10 +29,10 @@
 async def deploy(ops_test: OpsTest):
     """The app is up and running."""
     # Deploy temporal server, temporal admin and postgresql charms.
-    await ops_test.model.deploy(APP_NAME, channel="edge")
+    await ops_test.model.deploy(APP_NAME, channel="edge", config={"num-history-shards": 1})
     await ops_test.model.deploy(APP_NAME_ADMIN, channel="edge")
     await ops_test.model.deploy(APP_NAME_UI, channel="edge")
-    await ops_test.model.deploy("postgresql-k8s", channel="14", trust=True)
+    await ops_test.model.deploy("postgresql-k8s", channel="14/stable", trust=True)
 
     async with ops_test.fast_forward():
         await ops_test.model.wait_for_idle(
@@ -66,9 +66,6 @@ async def test_upgrade(self, ops_test: OpsTest):
         # This is to accmmodate for a self-resolving error which sometimes appears when Temporal
         # services attempt to connect to the cluster before the application is ready.
         await ops_test.model.applications[APP_NAME].refresh(path=str(charm), resources=resources)
-        await ops_test.model.applications[APP_NAME].set_config(
-            {"num-history-shards": "1"},
-        )
 
         await ops_test.model.wait_for_idle(
             apps=[APP_NAME], raise_on_error=False, status="active", raise_on_blocked=False, timeout=600

tox.ini

@@ -154,3 +154,16 @@ deps =
     -r{toxinidir}/requirements.txt
 commands =
     pytest {[vars]tst_path}integration/test_auth.py -v --tb native --ignore={[vars]tst_path}unit --log-cli-level=INFO -s {posargs} --destructive-mode
+
+[testenv:integration-server-upgrade]
+description = Run integration server upgrade testes
+deps =
+    ipdb==0.13.9
+    juju==3.3.0.0
+    pytest==7.1.3
+    pytest-operator==0.31.1
+    temporalio==1.1.0
+    pytest-asyncio==0.21
+    -r{toxinidir}/requirements.txt
+commands =
+    pytest {[vars]tst_path}integration/test_server_upgrade.py -v --tb native --ignore={[vars]tst_path}unit --log-cli-level=INFO -s {posargs} --destructive-mode