chore(slurmctld): use peer relation for ingress ip

charms/slurmctld/charmcraft.yaml

@@ -36,6 +36,10 @@ provides:
     interface: cos_agent
     limit: 1
 
+peers:
+  slurmctld-peer:
+    interface: slurmctld-peer
+
 assumes:
   - juju
 

charms/slurmctld/src/charm.py

@@ -9,7 +9,8 @@
 import subprocess
 from typing import Any, Dict, List, Optional, Union
 
-from constants import CHARM_MAINTAINED_SLURM_CONF_PARAMETERS
+from constants import CHARM_MAINTAINED_SLURM_CONF_PARAMETERS, PEER_RELATION
+from exceptions import IngressAddressUnavailableError
 from interface_slurmd import (
     PartitionAvailableEvent,
     PartitionUnavailableEvent,
@@ -303,7 +304,7 @@ def _assemble_slurmctld_parameters() -> dict[str, Any]:
 
         slurm_conf = SlurmConfig(
             ClusterName=self._cluster_name,
-            SlurmctldAddr=self._slurmd_ingress_address,
+            SlurmctldAddr=self._ingress_address,
             SlurmctldHost=[self._slurmctld.hostname],
             SlurmctldParameters=_assemble_slurmctld_parameters(),
             ProctrackType="proctrack/linuxproc" if is_container() else "proctrack/cgroup",
@@ -403,12 +404,13 @@ def hostname(self) -> str:
         return self._slurmctld.hostname
 
     @property
-    def _slurmd_ingress_address(self) -> str:
-        """Return the ingress_address from the slurmd relation if it exists."""
-        ingress_address = ""
-        if binding := self.model.get_binding("slurmd"):
-            ingress_address = f"{binding.network.ingress_address}"
-        return ingress_address
+    def _ingress_address(self) -> str:
+        """Return the ingress_address from the peer relation if it exists."""
+        if (peer_binding := self.model.get_binding(PEER_RELATION)) is not None:
+            ingress_address = f"{peer_binding.network.ingress_address}"
+            logger.debug(f"Slurmctld ingress_address: {ingress_address}")
+            return ingress_address
+        raise IngressAddressUnavailableError("Ingress address unavailable")
 
     @property
     def slurm_installed(self) -> bool:

charms/slurmctld/src/constants.py

@@ -3,6 +3,8 @@
 
 """This module provides constants for the slurmctld-operator charm."""
 
+PEER_RELATION = "slurmctld-peer"
+
 CHARM_MAINTAINED_SLURM_CONF_PARAMETERS = {
     "AuthAltParameters": {"jwt_key": "/var/lib/slurm/checkpoint/jwt_hs256.key"},
     "AuthAltTypes": ["auth/jwt"],

charms/slurmctld/src/exceptions.py

@@ -0,0 +1,13 @@
+# Copyright (c) 2024 Omnivector Corp
+# See LICENSE file for licensing details.
+
+"""Custom exceptions for the slurmctld operator."""
+
+
+class IngressAddressUnavailableError(Exception):
+    """Exception raised when a slurm operation failed."""
+
+    @property
+    def message(self) -> str:
+        """Return message passed as argument to exception."""
+        return self.args[0]

charms/slurmctld/tests/unit/test_charm.py

@@ -160,6 +160,7 @@ def test_on_slurmdbd_unavailable(self) -> None:
     def test_get_user_supplied_parameters(self, *_) -> None:
         """Test that user supplied parameters are parsed correctly."""
         self.harness.add_relation("slurmd", "slurmd")
+        self.harness.add_relation("slurmctld-peer", self.harness.charm.app.name)
         self.harness.update_config(
             {"slurm-conf-parameters": "JobAcctGatherFrequency=task=30,network=40"}
         )