feat: add method to check if a service is active

lib/charms/hpc_libs/v0/slurm_ops.py

@@ -62,6 +62,7 @@ def _on_install(self, _) -> None:
     "ConfigurationManager",
     "ServiceType",
     "SlurmManagerBase",
+    "SlurmOpsError",
 ]
 
 import json
@@ -131,7 +132,8 @@ def install() -> None:
 def version() -> str:
     """Get the current version of Slurm installed on the system."""
     info = yaml.safe_load(_snap("info", "slurm"))
-    ver: str = info["installed"]
+    if (ver := info.get("installed")) is None:
+        raise SlurmOpsError("unable to retrive snap info. Ensure slurm is correctly installed")
     return ver.split(maxsplit=1)[0]
 
 
@@ -208,6 +210,17 @@ def restart(self) -> None:
         """Restart service."""
         _snap("restart", f"slurm.{self._service.value}")
 
+    def active(self) -> bool:
+        """Return True if the service is active."""
+        info = yaml.safe_load(_snap("info", "slurm"))
+        if (services := info.get("services")) is None:
+            raise SlurmOpsError("unable to retrive snap info. Ensure slurm is correctly installed")
+
+        # Assume `services` contains the service, since `ServiceManager` is not exposed as a
+        # public interface for now.
+        # We don't do `"active" in state` because the word "active" is also part of "inactive" :)
+        return "inactive" not in services[f"slurm.{self._service.value}"]
+
 
 class ConfigurationManager:
     """Control configuration of a Slurm component."""

tests/integration/slurm_ops/test_manager.py

@@ -19,7 +19,6 @@ def slurmctld() -> SlurmManagerBase:
 def test_install(slurmctld: SlurmManagerBase) -> None:
     """Install Slurm using the manager."""
     slurm.install()
-    slurmctld.enable()
     slurmctld.munge.generate_key()
 
     with open("/var/snap/slurm/common/etc/munge/munge.key", "rb") as f:
@@ -40,7 +39,7 @@ def test_rotate_key(slurmctld: SlurmManagerBase) -> None:
 @pytest.mark.order(3)
 def test_slurm_config(slurmctld: SlurmManagerBase) -> None:
     """Test that the slurm config can be changed."""
-    slurmctld.config.set({"cluster-name": "test-cluster"})
+    slurmctld.config.set({"slurmctld-host": "test-slurm-ops", "cluster-name": "test-cluster"})
     value = slurmctld.config.get("cluster-name")
     assert value == "test-cluster"
 
@@ -57,6 +56,13 @@ def test_slurm_config(slurmctld: SlurmManagerBase) -> None:
 
 
 @pytest.mark.order(4)
+def test_enable_service(slurmctld: SlurmManagerBase) -> None:
+    """Test that the slurmctl daemon can be enabled."""
+    slurmctld.enable()
+    assert slurmctld.active()
+
+
+@pytest.mark.order(5)
 def test_version() -> None:
     """Test that the Slurm manager can report its version."""
     version = slurm.version()

tests/unit/test_slurm_ops.py

@@ -10,7 +10,7 @@
 from unittest.mock import patch
 
 import charms.hpc_libs.v0.slurm_ops as slurm
-from charms.hpc_libs.v0.slurm_ops import ServiceType, SlurmManagerBase
+from charms.hpc_libs.v0.slurm_ops import ServiceType, SlurmManagerBase, SlurmOpsError
 
 MUNGEKEY = b"1234567890"
 MUNGEKEY_BASE64 = base64.b64encode(MUNGEKEY)
@@ -27,15 +27,35 @@
     - slurm.command1
     - slurm.command2
 services:
+    slurm.logrotate:                 oneshot, enabled, inactive
     slurm.munged:                    simple, enabled, active
+    slurm.slurm-prometheus-exporter: simple, disabled, inactive
     slurm.slurmctld:                 simple, disabled, active
+    slurm.slurmd:                    simple, enabled, active
+    slurm.slurmdbd:                  simple, disabled, active
+    slurm.slurmrestd:                simple, disabled, active
 channels:
     latest/stable:    –
     latest/candidate: 23.11.7 2024-06-26 (460) 114MB classic
     latest/beta:      ↑
     latest/edge:      23.11.7 2024-06-26 (459) 114MB classic
 installed:          23.11.7             (x1) 114MB classic
 """
+SLURM_INFO_NOT_INSTALLED = """
+name:      slurm
+summary:   "Slurm: A Highly Scalable Workload Manager"
+publisher: –
+store-url: https://snapcraft.io/slurm
+license:   Apache-2.0
+description: |
+    Slurm is an open source, fault-tolerant, and highly scalable cluster
+    management and job scheduling system for large and small Linux clusters.
+channels:
+    latest/stable:    –
+    latest/candidate: 23.11.7 2024-06-26 (460) 114MB classic
+    latest/beta:      ↑
+    latest/edge:      23.11.7 2024-06-26 (459) 114MB classic
+"""
 
 
 @patch("charms.hpc_libs.v0.slurm_ops.subprocess.check_output")
@@ -61,12 +81,25 @@ def test_version(self, subcmd) -> None:
         self.assertEqual(args, ["snap", "info", "slurm"])
         self.assertEqual(version, "23.11.7")
 
+    def test_version_not_installed(self, subcmd) -> None:
+        """Test that `slurm_ops` throws when getting the installed version if the slurm snap is not installed."""
+        subcmd.return_value = SLURM_INFO_NOT_INSTALLED.encode()
+        with self.assertRaises(slurm.SlurmOpsError):
+            slurm.version()
+        args = subcmd.call_args[0][0]
+        self.assertEqual(args, ["snap", "info", "slurm"])
+
     def test_call_error(self, subcmd) -> None:
         """Test that `slurm_ops` propagates errors when a command fails."""
         subcmd.side_effect = subprocess.CalledProcessError(-1, cmd=[""], stderr=b"error")
         with self.assertRaises(slurm.SlurmOpsError):
             slurm.install()
 
+    def test_error_message(self, *_) -> None:
+        """Test that `SlurmOpsError` stores the correct message."""
+        message = "error message!"
+        self.assertEqual(SlurmOpsError(message).message, message)
+
 
 @patch("charms.hpc_libs.v0.slurm_ops.subprocess.check_output")
 class SlurmOpsBase:
@@ -101,6 +134,19 @@ def test_restart(self, subcmd, *_) -> None:
         args = subcmd.call_args[0][0]
         self.assertEqual(args, ["snap", "restart", f"slurm.{self.manager._service.value}"])
 
+    def test_active(self, subcmd, *_) -> None:
+        """Test that the manager can detect that a service is active."""
+        subcmd.return_value = SLURM_INFO.encode()
+        self.assertTrue(self.manager.active())
+
+    def test_active_not_installed(self, subcmd, *_) -> None:
+        """Test that the manager throws an error when calling `active` if the snap is not installed."""
+        subcmd.return_value = SLURM_INFO_NOT_INSTALLED.encode()
+        with self.assertRaises(slurm.SlurmOpsError):
+            self.manager.active()
+        args = subcmd.call_args[0][0]
+        self.assertEqual(args, ["snap", "info", "slurm"])
+
     def test_get_options(self, subcmd) -> None:
         """Test that the manager correctly collects all requested configuration options."""
         subcmd.return_value = '{"%(name)s.key1": "value1", "%(name)s.key2": "value2"}' % {