Fix slurm exit-code detection

equinor · Dec 3, 2024 · 1d0de82 · 1d0de82
1 parent 309a535
commit 1d0de82
Showing 1 changed file with 10 additions and 7 deletions.
diff --git a/src/ert/scheduler/slurm_driver.py b/src/ert/scheduler/slurm_driver.py
@@ -368,9 +368,13 @@ async def _poll_once_by_scontrol(
             return self._scontrol_cache[missing_job_id]
 
         process = await asyncio.create_subprocess_exec(
-            self._scontrol,
-            "show",
-            "job",
+            "sacct",
+            "-X",
+            "-n",
+            " -o",
+            "State,ExitCode",
+            "-P",
+            "-j",
             str(missing_job_id),
             stdout=asyncio.subprocess.PIPE,
             stderr=asyncio.subprocess.PIPE,
@@ -441,9 +445,8 @@ def _seconds_to_slurm_time_format(seconds: float) -> str:
 
 
 def _parse_scontrol_output(output: str) -> ScontrolInfo:
-    values = dict(w.split("=", 1) for w in output.split())
-    exit_code_str = values.get("ExitCode")
+    state_str, exit_code_str = output.split("|")
     exit_code = None
     if exit_code_str:
-        exit_code = int(exit_code_str.split(":")[0])
-    return ScontrolInfo(JobStatus[values["JobState"]], exit_code)
+        exit_code = int(output.split(":")[0])
+    return ScontrolInfo(JobStatus[state_str], exit_code)