Merge pull request #469 from omnivector-solutions/tucker/PENG-2064--a…

…dd-slurm-job-state

PENG-2064 Capture job state info in jobbergate

examples/simple-job-script.py

@@ -8,5 +8,5 @@
 
 print("Executing simple job script")
 
-output_file = Path(f"/nfs/simple-output.txt")
+output_file = Path(f"simple-output.txt")
 output_file.write_text(f"Simple output from {gethostname()}")

examples/simple-job-script.sh

@@ -0,0 +1,8 @@
+#!/bin/bash
+
+#SBATCH -J dummy_job
+#SBATCH -t 60
+
+echo "Executing simple job script"
+
+echo "Simple output from $HOSTNAME" > simple-output.txt

examples/sleepy-job-script.sh

@@ -0,0 +1,8 @@
+#!/bin/bash
+
+#SBATCH -J dummy_job
+#SBATCH -t 0:0:60
+
+echo "Executing sleepy job script"
+sleep 120
+echo "Sleepy job script woke up"

jobbergate-agent/CHANGELOG.md

@@ -3,7 +3,12 @@
 This file keeps track of all notable changes to jobbergate-core
 
 ## Unreleased
-
+- Revamped job_submissions statuses and tracked more details slurm job_state [PENG-2064]
+  - Renamed the finish.py module to update.py
+  - Changed logic such that all active jobs have their job state updated on each pass
+  - Moved all logic from the api.py module to update.py and submit.py
+  - Updated schemas for revised job payloads
+  - Updated and added unit tests
 
 ## 4.3.0a6 -- 2024-02-06
 ## 4.3.0a5 -- 2024-02-02

jobbergate-agent/Dockerfile.dev

@@ -1,4 +1,4 @@
-FROM python:3.8-slim-buster
+FROM python:3.12-slim-bookworm
 
 WORKDIR /app
 

jobbergate-agent/jobbergate_agent/jobbergate/constants.py

@@ -1,31 +1,8 @@
-from collections import defaultdict
-from enum import Enum
-from typing import DefaultDict
+from auto_name_enum import AutoNameEnum, auto
 
 
-class FileType(str, Enum):
+class FileType(AutoNameEnum):
     """File type enum."""
 
-    ENTRYPOINT = "ENTRYPOINT"
-    SUPPORT = "SUPPORT"
-
-
-class JobSubmissionStatus(str, Enum):
-    """
-    Enumeration of possible job_submission statuses.
-    """
-
-    CREATED = "CREATED"
-    SUBMITTED = "SUBMITTED"
-    COMPLETED = "COMPLETED"
-    FAILED = "FAILED"
-    CANCELLED = "CANCELLED"
-    REJECTED = "REJECTED"
-
-
-status_map: DefaultDict[str, JobSubmissionStatus] = defaultdict(
-    lambda: JobSubmissionStatus.SUBMITTED,
-    COMPLETED=JobSubmissionStatus.COMPLETED,
-    FAILED=JobSubmissionStatus.FAILED,
-    CANCELLED=JobSubmissionStatus.CANCELLED,
-)
+    ENTRYPOINT = auto()
+    SUPPORT = auto()

jobbergate-agent/jobbergate_agent/jobbergate/schemas.py

@@ -3,7 +3,7 @@
 
 import pydantic
 
-from jobbergate_agent.jobbergate.constants import FileType, JobSubmissionStatus, status_map
+from jobbergate_agent.jobbergate.constants import FileType
 
 
 class JobScriptFile(pydantic.BaseModel, extra=pydantic.Extra.ignore):
@@ -94,17 +94,13 @@ class SlurmSubmitResponse(pydantic.BaseModel, extra=pydantic.Extra.ignore):
     job_id: Optional[int]
 
 
-class SlurmSubmittedJobStatus(pydantic.BaseModel, extra=pydantic.Extra.ignore):
+class SlurmJobData(pydantic.BaseModel, extra=pydantic.Extra.ignore):
     """
-    Specialized model for the cluster-agent to pull a concluded job_submission.
+    Specialized model for the cluster-agent to pull job state information from slurm and post the data as an update
+    to the Jobbergate API.
     """
 
     job_id: Optional[int]
     job_state: Optional[str]
+    job_info: Optional[str]
     state_reason: Optional[str]
-
-    @property
-    def jobbergate_status(self) -> Optional[JobSubmissionStatus]:
-        if self.job_state:
-            return status_map[self.job_state]
-        return None