feat!: overhaul slurmctld API part 2

Merge pull request #27 from jamesbeedy/small_fixes

Summary of Changes:

- remove fluentbit integration and associated code
- adds back creation of the the state dir
- fix path to jwt_key

charmcraft.yaml

@@ -29,8 +29,6 @@ requires:
     interface: slurmdbd
   slurmrestd:
     interface: slurmrestd
-  fluentbit:
-    interface: fluentbit
 
 assumes:
   - juju

src/charm.py

@@ -9,8 +9,7 @@
 import subprocess
 from typing import Any, Dict, List, Optional, Union
 
-from charms.fluentbit.v0.fluentbit import FluentbitClient
-from constants import CHARM_MAINTAINED_SLURM_CONF_PARAMETERS, FLUENTBIT_CONFIG, SLURM_CONF_PATH
+from constants import CHARM_MAINTAINED_SLURM_CONF_PARAMETERS, SLURM_CONF_PATH
 from interface_slurmd import (
     PartitionAvailableEvent,
     PartitionUnavailableEvent,
@@ -34,7 +33,6 @@
     CharmBase,
     ConfigChangedEvent,
     InstallEvent,
-    RelationCreatedEvent,
     StoredState,
     UpdateStatusEvent,
     WaitingStatus,
@@ -68,29 +66,21 @@ def __init__(self, *args):
 
         self._slurmctld_manager = SlurmctldManager()
 
-        self._fluentbit = FluentbitClient(self, "fluentbit")
         self._slurmd = Slurmd(self, "slurmd")
         self._slurmdbd = Slurmdbd(self, "slurmdbd")
         self._slurmrestd = Slurmrestd(self, "slurmrestd")
 
         event_handler_bindings = {
-            # Charm lifecycle hook events
             self.on.install: self._on_install,
             self.on.update_status: self._on_update_status,
             self.on.config_changed: self._on_config_changed,
-            # slurmdbd lifecycle hook events
             self._slurmdbd.on.slurmdbd_available: self._on_slurmdbd_available,
             self._slurmdbd.on.slurmdbd_unavailable: self._on_slurmdbd_unavailable,
-            # slurmd lifecycle hook events
             self._slurmd.on.partition_available: self._on_write_slurm_conf,
             self._slurmd.on.partition_unavailable: self._on_write_slurm_conf,
             self._slurmd.on.slurmd_available: self._on_write_slurm_conf,
             self._slurmd.on.slurmd_departed: self._on_write_slurm_conf,
-            # slurmrestd available
             self._slurmrestd.on.slurmrestd_available: self._on_slurmrestd_available,
-            # fluentbit
-            self.on["fluentbit"].relation_created: self._on_fluentbit_relation_created,
-            # actions
             self.on.show_current_config_action: self._on_show_current_config_action,
             self.on.drain_action: self._on_drain_nodes_action,
             self.on.resume_action: self._on_resume_nodes_action,
@@ -173,11 +163,6 @@ def _on_show_current_config_action(self, event: ActionEvent) -> None:
         slurm_conf = SLURM_CONF_PATH.read_text()
         event.set_results({"slurm.conf": slurm_conf})
 
-    def _on_fluentbit_relation_created(self, event: RelationCreatedEvent) -> None:
-        """Set up Fluentbit log forwarding."""
-        logger.debug("## Configuring fluentbit")
-        self._fluentbit.configure(FLUENTBIT_CONFIG)
-
     def _on_slurmrestd_available(self, event: SlurmrestdAvailableEvent) -> None:
         """Check that we have slurm_config when slurmrestd available otherwise defer the event."""
         if self.model.unit.is_leader():

src/constants.py

@@ -8,7 +8,7 @@
 SLURM_GROUP = "slurm"
 
 CHARM_MAINTAINED_SLURM_CONF_PARAMETERS = {
-    "AuthAltParameters": "jwt_key=/var/spool/slurmctldjwt_hs256.key",
+    "AuthAltParameters": "jwt_key=/var/spool/slurmctld/jwt_hs256.key",
     "AuthAltTypes": "auth/jwt",
     "AuthInfo": "/var/run/munge/munge.socket.2",
     "AuthType": "auth/munge",
@@ -34,37 +34,6 @@
     "RebootProgram": '"/usr/sbin/reboot --reboot"',
 }
 
-
-FLUENTBIT_CONFIG = [
-    {
-        "input": [
-            ("name", "tail"),
-            ("path", "/var/log/slurm/slurmctld.log"),
-            ("path_key", "filename"),
-            ("tag", "slurmctld"),
-            ("parser", "slurm"),
-        ]
-    },
-    {
-        "parser": [
-            ("name", "slurm"),
-            ("format", "regex"),
-            ("regex", r"^\[(?<time>[^\]]*)\] (?<log>.*)$"),
-            ("time_key", "time"),
-            ("time_format", "%Y-%m-%dT%H:%M:%S.%L"),
-        ]
-    },
-    {
-        "filter": [
-            ("name", "record_modifier"),
-            ("match", "slurmctld"),
-            ("record", "hostname ${HOSTNAME}"),
-            ("record", "service slurmctld"),
-        ]
-    },
-]
-
-
 UBUNTU_HPC_PPA_KEY = """
 -----BEGIN PGP PUBLIC KEY BLOCK-----
 Comment: Hostname:

src/slurmctld_ops.py

@@ -138,6 +138,13 @@ def install(self) -> bool:
             return False
         systemd.service_stop("munge")
 
+        spool_dir = Path("/var/spool/slurmctld")
+        if not spool_dir.exists():
+            spool_dir.mkdir()
+
+        slurm_user_uid, slurm_group_gid = _get_slurm_user_uid_and_slurm_group_gid()
+        os.chown(f"{spool_dir}", slurm_user_uid, slurm_group_gid)
+
         return True
 
     def version(self) -> str: