chore: fix CI pipeline

A fix was applied because there was a bug on the way we deferred `_on_slurmctld_available`.

.github/workflows/ci.yaml

@@ -66,7 +66,7 @@ jobs:
     strategy:
       fail-fast: true
       matrix:
-        bases:          
+        bases:
           - ubuntu@22.04
     name: Integration tests (LXD) | ${{ matrix.bases }}
     runs-on: ubuntu-latest
@@ -78,10 +78,17 @@ jobs:
     steps:
       - name: Checkout
         uses: actions/checkout@v4
+        with:
+          path: main
+      - name: Fetch slurmctld
+        uses: actions/checkout@v4
+        with:
+          repository: charmed-hpc/slurmctld-operator
+          path: slurmctld-operator
       - name: Setup operator environment
         uses: charmed-kubernetes/actions-operator@main
         with:
           provider: lxd
-          juju-channel: 3.1/stable
+          juju-channel: 3.4/stable
       - name: Run tests
-        run: tox run -e integration -- --charm-base=${{ matrix.bases }}
+        run: cd main && tox run -e integration -- --charm-base=${{ matrix.bases }} --use-local

.github/workflows/release.yaml

@@ -32,10 +32,10 @@ jobs:
       - name: Checkout
         uses: actions/checkout@v3
       - name: Select charmhub channel
-        uses: canonical/charming-actions/channel@2.2.0
+        uses: canonical/charming-actions/channel@2.5.0-rc
         id: channel
       - name: Upload charm to charmhub
-        uses: canonical/charming-actions/upload-charm@2.2.0
+        uses: canonical/charming-actions/upload-charm@2.5.0-rc
         with:
           credentials: "${{ secrets.CHARMCRAFT_AUTH }}"
           github-token: "${{ secrets.GITHUB_TOKEN }}"

src/charm.py

@@ -102,7 +102,12 @@ def _on_slurmctld_available(self, event: SlurmctldAvailableEvent) -> None:
             self._slurmdbd_ops_manager.write_munge_key(munge_key)
             self._slurmdbd_ops_manager.start_munge()
 
-        self._write_config_and_restart_slurmdbd(event)
+        # Don't try to write the config before the database has been created.
+        # Otherwise, this will trigger a defer on this event, which we don't really need
+        # or the munge service will restart too many times, triggering a restart limit on
+        # systemctl.
+        if self._stored.db_info:
+            self._write_config_and_restart_slurmdbd(event)
 
     def _on_database_created(self, event: DatabaseCreatedEvent) -> None:
         """Process the DatabaseCreatedEvent and updates the database parameters.

tests/integration/test_charm.py

@@ -71,7 +71,7 @@ async def test_build_and_deploy_against_edge(
         ),
     )
     # Set relations for charmed applications.
-    await ops_test.model.integrate(f"{SLURMDBD}:{SLURMDBD}", f"{SLURMCTLD}:{SLURMDBD}")
+    await ops_test.model.integrate(f"{SLURMDBD}:{SLURMCTLD}", f"{SLURMCTLD}:{SLURMDBD}")
     await ops_test.model.integrate(f"{ROUTER}:backend-database", f"{DATABASE}:database")
     await ops_test.model.integrate(f"{SLURMDBD}:database", f"{ROUTER}:database")
     # Reduce the update status frequency to accelerate the triggering of deferred events.

tox.ini

@@ -10,7 +10,7 @@ env_list = lint, type, unit
 [vars]
 src_path = {toxinidir}/src
 tst_path = {toxinidir}/tests
-all_path = {[vars]src_path} {[vars]tst_path} 
+all_path = {[vars]src_path} {[vars]tst_path}
 
 [testenv]
 setenv =