update based on obs team comments

.gitignore

@@ -1,3 +1,4 @@
 *.charm
 build/
 __pycache__/
+.tox

metadata.yaml

@@ -19,7 +19,7 @@ provides:
     interface: object-storage
     schema: https://raw.githubusercontent.com/canonical/operator-schemas/master/object-storage.yaml
     versions: [v1]
-  monitoring:
+  metrics-endpoint:
     interface: prometheus_scrape
   grafana-dashboard:
     interface: grafana_dashboard

src/charm.py

@@ -34,7 +34,6 @@ def __init__(self, *args):
 
         self.prometheus_provider = MetricsEndpointProvider(
             charm=self,
-            relation_name="monitoring",
             jobs=[
                 {
                     "job_name": "minio_metrics",
@@ -54,9 +53,6 @@ def __init__(self, *args):
             self.on.upgrade_charm,
             self.on["object-storage"].relation_changed,
             self.on["object-storage"].relation_joined,
-            self.on["monitoring"].relation_changed,
-            self.on["monitoring"].relation_broken,
-            self.on["monitoring"].relation_departed,
         ]:
             self.framework.observe(event, self.main)
 
@@ -197,14 +193,14 @@ def _gen_pass() -> str:
 
 
 class CheckFailed(Exception):
-    """ Raise this exception if one of the checks in main fails. """
+    """Raise this exception if one of the checks in main fails."""
 
     def __init__(self, msg, status_type=None):
         super().__init__()
 
-        self.msg = msg
+        self.msg = str(msg)
         self.status_type = status_type
-        self.status = status_type(msg)
+        self.status = status_type(self.msg)
 
 
 if __name__ == "__main__":

src/prometheus_alert_rules/unit_unavailable.rule

@@ -0,0 +1,10 @@
+alert: MinioUnitIsUnavailable
+expr: up < 1
+for: 0m
+labels:
+  severity: critical
+annotations:
+  summary: Minio unit {{ $labels.juju_model }}/{{ $labels.juju_unit }} unavailable
+  description: >
+    The Minio unit {{ $labels.juju_model }} {{ $labels.juju_unit }} is unavailable
+    LABELS = {{ $labels }}

test-requirements.txt

@@ -4,5 +4,6 @@
 pytest
 pyyaml
 flake8
-black==20.8b1
+black
 flake8-copyright<0.3
+requests

tests/integration/test_charm.py

@@ -6,6 +6,8 @@
 
 import pytest
 import yaml
+import requests
+import json
 from pytest_operator.plugin import OpsTest
 
 log = logging.getLogger(__name__)
@@ -19,6 +21,9 @@
 
 APP_NAME = "minio"
 CHARM_ROOT = "."
+PROMETHEUS = "prometheus-k8s"
+GRAFANA = "grafana-k8s"
+PROMETHEUS_SCRAPE = "prometheus-scrape-config-k8s"
 
 
 @pytest.mark.abort_on_fail
@@ -118,3 +123,41 @@ async def test_connect_to_console(ops_test: OpsTest):
     assert (
         ret_code == 0
     ), f"Test returned code {ret_code} with stdout:\n{stdout}\nstderr:\n{stderr}"
+
+
+async def test_deploy_with_prometheus_and_grafana(ops_test):
+    scrape_config = {"scrape_interval": "30s"}
+    await ops_test.model.deploy(PROMETHEUS, channel="latest/beta")
+    await ops_test.model.deploy(GRAFANA, channel="latest/beta")
+    await ops_test.model.deploy(
+        PROMETHEUS_SCRAPE, channel="latest/beta", config=scrape_config
+    )
+    await ops_test.model.add_relation(APP_NAME, PROMETHEUS_SCRAPE)
+    await ops_test.model.add_relation(PROMETHEUS, PROMETHEUS_SCRAPE)
+    await ops_test.model.add_relation(PROMETHEUS, GRAFANA)
+    await ops_test.model.add_relation(APP_NAME, GRAFANA)
+
+    await ops_test.model.wait_for_idle(
+        [APP_NAME, PROMETHEUS, GRAFANA, PROMETHEUS_SCRAPE], status="active"
+    )
+
+
+async def test_correct_observability_setup(ops_test):
+    status = await ops_test.model.get_status()
+    prometheus_unit_ip = status["applications"][PROMETHEUS]["units"][f"{PROMETHEUS}/0"][
+        "address"
+    ]
+    r = requests.get(
+        f'http://{prometheus_unit_ip}:9090/api/v1/query?query=up{{juju_application="{APP_NAME}"}}'
+    )
+    response = json.loads(r.content.decode("utf-8"))
+    assert response["status"] == "success"
+    assert len(response["data"]["result"]) == len(
+        ops_test.model.applications[APP_NAME].units
+    )
+
+    response_metric = response["data"]["result"][0]["metric"]
+    assert response_metric["juju_application"] == APP_NAME
+    assert response_metric["juju_charm"] == APP_NAME
+    assert response_metric["juju_model"] == ops_test.model_name
+    assert response_metric["juju_unit"] == f"{APP_NAME}/0"

tox.ini

@@ -29,5 +29,5 @@ commands = pytest -vv {toxinidir}/tests/unit
 [testenv:lint]
 commands =
     flake8 {toxinidir}/src {toxinidir}/tests
-    black --check {toxinidir}/src {toxinidir}/tests
+    black --check --diff {toxinidir}/src {toxinidir}/tests