test: ssh-debug integration test (#193)

* feat: initial ssh-debug integration

* fix: linting and test fixes

* test: add debug info tests

* fix: relation data & tests

* test: fix unit test

* feat: flush runners on relation changed

* fix: lint

* test: validate SHA fingerprint format

* fix: flush only idle runners

* fix: remove remote unit filter

* fix: use underscore for env vars

* test: tmate action test

* test: tmate action test

* test: add ssh_debug test to workflow

* test: rename relation

* fix: json serializable state

* fix: optional state serialization

* fix: use machine address

* fix: test branch naming

* test: fix iterator error

* test: trigger w/ longer timeout

* test: trigger

* chore: move test dependencies

* test: refactor tests

* test: refactor tests

* fix: lint

* fix: lint

* test: move imports helpers - charm_metrics_helpers - modules

* Update tests/integration/helpers.py

Co-authored-by: Christopher Bartz <[email protected]>

* test: lint fix

---------

Co-authored-by: Christopher Bartz <[email protected]>

.github/workflows/integration_test.yaml

@@ -13,7 +13,7 @@ jobs:
       pre-run-script: scripts/pre-integration-test.sh
       provider: lxd
       test-tox-env: integration-juju2.9
-      modules: '["test_charm_fork_repo", "test_charm_no_runner", "test_charm_scheduled_events", "test_charm_one_runner", "test_charm_metrics_success", "test_charm_metrics_failure", "test_self_hosted_runner", "test_charm_with_proxy", "test_charm_with_juju_storage"]'
+      modules: '["test_charm_fork_repo", "test_charm_no_runner", "test_charm_scheduled_events", "test_charm_one_runner", "test_charm_metrics_success", "test_charm_metrics_failure", "test_self_hosted_runner", "test_charm_with_proxy", "test_charm_with_juju_storage", "test_debug_ssh"]'
   integration-tests-juju3:
     name: Integration test with juju 3.1
     uses: canonical/operator-workflows/.github/workflows/integration_test.yaml@main
@@ -23,4 +23,4 @@ jobs:
       pre-run-script: scripts/pre-integration-test.sh
       provider: lxd
       test-tox-env: integration-juju3.1
-      modules: '["test_charm_fork_repo", "test_charm_no_runner", "test_charm_scheduled_events", "test_charm_one_runner", "test_charm_metrics_success", "test_charm_metrics_failure", "test_self_hosted_runner", "test_charm_with_proxy", "test_charm_with_juju_storage"]'
+      modules: '["test_charm_fork_repo", "test_charm_no_runner", "test_charm_scheduled_events", "test_charm_one_runner", "test_charm_metrics_success", "test_charm_metrics_failure", "test_self_hosted_runner", "test_charm_with_proxy", "test_charm_with_juju_storage", "test_debug_ssh"]'

src/charm.py

@@ -932,6 +932,7 @@ def _on_debug_ssh_relation_changed(self, _: ops.RelationChangedEvent) -> None:
         """Handle debug ssh relation changed event."""
         runner_manager = self._get_runner_manager()
         runner_manager.flush(flush_busy=False)
+        self._reconcile_runners(runner_manager)
 
 
 if __name__ == "__main__":

src/charm_state.py

@@ -309,6 +309,9 @@ def from_charm(cls, charm: CharmBase) -> "State":
         # Convert pydantic object to python object serializable by json module.
         state_dict["proxy_config"] = json.loads(state_dict["proxy_config"].json())
         state_dict["charm_config"] = json.loads(state_dict["charm_config"].json())
+        state_dict["ssh_debug_info"] = (
+            json.loads(state_dict["ssh_debug_info"].json()) if ssh_debug_info else None
+        )
         json_data = json.dumps(state_dict, ensure_ascii=False)
         CHARM_STATE_PATH.write_text(json_data, encoding="utf-8")
 

tests/integration/charm_metrics_helpers.py

@@ -6,11 +6,9 @@
 
 import json
 import logging
-from datetime import datetime, timezone
 from time import sleep
 
 import requests
-from github.Branch import Branch
 from github.Repository import Repository
 from github.Workflow import Workflow
 from juju.application import Application
@@ -19,76 +17,13 @@
 from github_type import JobConclusion
 from metrics import METRICS_LOG_PATH
 from runner_metrics import PostJobStatus
-from tests.integration.helpers import (
-    DISPATCH_FAILURE_TEST_WORKFLOW_FILENAME,
-    DISPATCH_TEST_WORKFLOW_FILENAME,
-    get_runner_name,
-    get_runner_names,
-    run_in_unit,
-)
+from tests.integration.helpers import JOB_LOG_START_MSG_TEMPLATE, get_runner_name, run_in_unit
 
 TEST_WORKFLOW_NAMES = [
     "Workflow Dispatch Tests",
     "Workflow Dispatch Crash Tests",
     "Workflow Dispatch Failure Tests 2a34f8b1-41e4-4bcb-9bbf-7a74e6c482f7",
 ]
-JOB_LOG_START_MSG_TEMPLATE = "Job is about to start running on the runner: {runner_name}"
-
-
-async def _wait_until_runner_is_used_up(runner_name: str, unit: Unit):
-    """Wait until the runner is used up.
-
-    Args:
-        runner_name: The runner name to wait for.
-        unit: The unit which contains the runner.
-    """
-    for _ in range(30):
-        runners = await get_runner_names(unit)
-        if runner_name not in runners:
-            break
-        sleep(30)
-    else:
-        assert False, "Timeout while waiting for the runner to be used up"
-
-
-async def _assert_workflow_run_conclusion(
-    runner_name: str, conclusion: str, workflow: Workflow, start_time: datetime
-):
-    """Assert that the workflow run has the expected conclusion.
-
-    Args:
-        runner_name: The runner name to assert the workflow run conclusion for.
-        conclusion: The expected workflow run conclusion.
-        workflow: The workflow to assert the workflow run conclusion for.
-        start_time: The start time of the workflow.
-    """
-    for run in workflow.get_runs(created=f">={start_time.isoformat()}"):
-        logs_url = run.jobs()[0].logs_url()
-        logs = requests.get(logs_url).content.decode("utf-8")
-
-        if JOB_LOG_START_MSG_TEMPLATE.format(runner_name=runner_name) in logs:
-            assert run.jobs()[0].conclusion == conclusion
-
-
-async def _wait_for_workflow_to_complete(
-    unit: Unit, workflow: Workflow, conclusion: str, start_time: datetime
-):
-    """Wait for the workflow to complete.
-
-    Args:
-        unit: The unit which contains the runner.
-        workflow: The workflow to wait for.
-        conclusion: The workflow conclusion to wait for.
-        start_time: The start time of the workflow.
-    """
-    runner_name = await get_runner_name(unit)
-    await _wait_until_runner_is_used_up(runner_name, unit)
-    # Wait for the workflow log to contain the conclusion
-    sleep(60)
-
-    await _assert_workflow_run_conclusion(
-        runner_name=runner_name, conclusion=conclusion, workflow=workflow, start_time=start_time
-    )
 
 
 async def _wait_for_workflow_to_start(unit: Unit, workflow: Workflow):
@@ -189,34 +124,6 @@ async def _cancel_workflow_run(unit: Unit, workflow: Workflow):
                 run.cancel()
 
 
-async def dispatch_workflow(
-    app: Application, branch: Branch, github_repository: Repository, conclusion: str
-):
-    """Dispatch a workflow on a branch for the runner to run.
-
-    The function assumes that there is only one runner running in the unit.
-
-    Args:
-        app: The charm to dispatch the workflow for.
-        branch: The branch to dispatch the workflow on.
-        github_repository: The github repository to dispatch the workflow on.
-        conclusion: The expected workflow run conclusion.
-    """
-    start_time = datetime.now(timezone.utc)
-
-    workflow = github_repository.get_workflow(id_or_file_name=DISPATCH_TEST_WORKFLOW_FILENAME)
-    if conclusion == "failure":
-        workflow = github_repository.get_workflow(
-            id_or_file_name=DISPATCH_FAILURE_TEST_WORKFLOW_FILENAME
-        )
-
-    # The `create_dispatch` returns True on success.
-    assert workflow.create_dispatch(branch, {"runner": app.name})
-    await _wait_for_workflow_to_complete(
-        unit=app.units[0], workflow=workflow, conclusion=conclusion, start_time=start_time
-    )
-
-
 async def assert_events_after_reconciliation(
     app: Application, github_repository: Repository, post_job_status: PostJobStatus
 ):

tests/integration/conftest.py

@@ -12,17 +12,20 @@
 import pytest
 import pytest_asyncio
 import yaml
+from git import Repo
 from github import Github, GithubException
 from github.Branch import Branch
 from github.Repository import Repository
 from juju.application import Application
+from juju.client._definitions import FullStatus, UnitStatus
 from juju.model import Model
 from pytest_operator.plugin import OpsTest
 
 from tests.integration.helpers import (
     deploy_github_runner_charm,
     ensure_charm_has_runner,
     reconcile,
+    wait_for,
 )
 from tests.status_name import ACTIVE
 
@@ -246,6 +249,64 @@ async def app_runner(
     return application
 
 
+@pytest_asyncio.fixture(scope="module", name="app_no_wait")
+async def app_no_wait_fixture(
+    model: Model,
+    charm_file: str,
+    app_name: str,
+    path: str,
+    token: str,
+    http_proxy: str,
+    https_proxy: str,
+    no_proxy: str,
+) -> AsyncIterator[Application]:
+    """GitHub runner charm application without waiting for active."""
+    app: Application = await deploy_github_runner_charm(
+        model=model,
+        charm_file=charm_file,
+        app_name=app_name,
+        path=path,
+        token=token,
+        runner_storage="memory",
+        http_proxy=http_proxy,
+        https_proxy=https_proxy,
+        no_proxy=no_proxy,
+        reconcile_interval=60,
+        wait_idle=False,
+    )
+    await app.set_config({"virtual-machines": "1"})
+    return app
+
+
+@pytest_asyncio.fixture(scope="module", name="tmate_ssh_server_app")
+async def tmate_ssh_server_app_fixture(
+    model: Model, app_no_wait: Application
+) -> AsyncIterator[Application]:
+    """tmate-ssh-server charm application related to GitHub-Runner app charm."""
+    tmate_app: Application = await model.deploy("tmate-ssh-server", channel="edge")
+    await app_no_wait.relate("debug-ssh", f"{tmate_app.name}:debug-ssh")
+    await model.wait_for_idle(status=ACTIVE, timeout=60 * 30)
+
+    return tmate_app
+
+
+@pytest_asyncio.fixture(scope="module", name="tmate_ssh_server_unit_ip")
+async def tmate_ssh_server_unit_ip_fixture(
+    model: Model,
+    tmate_ssh_server_app: Application,
+) -> AsyncIterator[str]:
+    """tmate-ssh-server charm unit ip."""
+    status: FullStatus = await model.get_status([tmate_ssh_server_app.name])
+    try:
+        unit_status: UnitStatus = next(
+            iter(status.applications[tmate_ssh_server_app.name].units.values())
+        )
+        assert unit_status.public_address, "Invalid unit address"
+        return unit_status.public_address
+    except StopIteration as exc:
+        raise StopIteration("Invalid unit status") from exc
+
+
 @pytest.fixture(scope="module")
 def github_client(token: str) -> Github:
     """Returns the github client."""
@@ -357,6 +418,31 @@ async def app_juju_storage(
     return application
 
 
+@pytest_asyncio.fixture(scope="module", name="test_github_branch")
+async def test_github_branch_fixture(github_repository: Repository) -> AsyncIterator[Branch]:
+    """Create a new branch for testing, from latest commit in current branch."""
+    test_branch = f"test-{secrets.token_hex(4)}"
+    branch_ref = github_repository.create_git_ref(
+        ref=f"refs/heads/{test_branch}", sha=Repo().head.commit.hexsha
+    )
+
+    def get_branch():
+        """Get newly created branch."""
+        try:
+            branch = github_repository.get_branch(test_branch)
+        except GithubException as err:
+            if err.status == 404:
+                return False
+            raise
+        return branch
+
+    await wait_for(get_branch)
+
+    yield get_branch()
+
+    branch_ref.delete()
+
+
 @pytest_asyncio.fixture(scope="module", name="app_with_grafana_agent")
 async def app_with_grafana_agent_integrated_fixture(
     model: Model, app_no_runner: Application