Merge branch 'main' into nemo2.0

NVIDIA · Nov 20, 2024 · 9ac9520 · 9ac9520
2 parents 12122d5 + e6055fe
commit 9ac9520
Show file tree

Hide file tree

Showing 2 changed files with 6 additions and 2 deletions.
diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
@@ -100,3 +100,7 @@ jobs:
           # this checks that all TOMLs are valid, Test Scenarios are checked agains _only_ the tests in the specified directory
           cloudai verify-configs --tests-dir conf/common/test conf/common
           cloudai verify-configs --tests-dir conf/release/spcx/l40s/test conf/release/spcx/l40s
+
+          cloudai dry-run --system-config conf/common/system/example_slurm_cluster.toml \
+            --tests-dir conf/common/test/ \
+            --test-scenario conf/common/test_scenario/ucc_test.toml
diff --git a/src/cloudai/_core/base_runner.py b/src/cloudai/_core/base_runner.py
@@ -204,7 +204,7 @@ async def check_start_post_init_dependencies(self):
         items = list(self.testrun_to_job_map.items())
 
         for tr, job in items:
-            if self.system.is_job_running(job):
+            if self.mode == "dry-run" or self.system.is_job_running(job):
                 await self.check_and_schedule_start_post_init_dependent_tests(tr)
 
     async def check_and_schedule_start_post_init_dependent_tests(self, started_test_run: TestRun):
@@ -279,7 +279,7 @@ async def monitor_jobs(self) -> int:
         successful_jobs_count = 0
 
         for job in list(self.jobs):
-            if self.system.is_job_completed(job):
+            if self.mode == "dry-run" or self.system.is_job_completed(job):
                 await self.job_completion_callback(job)
 
                 if self.mode == "dry-run":