Merge branch 'RESTAPI-1150-polling-output' into 'master'

Return empty `job_data_out` and `job_file_err` while job is pending

See merge request firecrest/firecrest!308

CHANGELOG.md

@@ -22,6 +22,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 ### Fixed
 
 - Fix parsing in `GET /utilities/ls` endpoint.
+- The job fields `job_data_out` and `job_file_err` from `GET /compute/jobs` will be empty for jobs that are still pending (so that there is no confusion with older output/error files).
 
 ## [1.16.0]
 

src/common/schedulers/__init__.py

@@ -94,6 +94,7 @@ def job_info(self, jobid):
         * Job output filename
         * Job error filename
         * Job script filename
+        * Job state
         """
         pass
 
@@ -104,6 +105,7 @@ def parse_job_info(self, output):
         * Job output filename
         * Job error filename
         * Job script filename
+        * Job state
         """
         pass
 
@@ -202,3 +204,9 @@ def check_job_time(self, job_time):
         * DD-HH DD-HH:MM
         * DD-HH:MM:SS
         """
+
+    @abc.abstractmethod
+    def job_is_pending(self, state):
+        """Check if the job is pending based on the state.
+        """
+        pass

src/common/schedulers/slurm.py

@@ -554,3 +554,15 @@ def check_job_time(self, job_time):
             logger.error(e, exc_info=True)
 
             return False
+
+    def job_is_pending(self, state):
+        # Complete list in Slurm's documentation:
+        # https://slurm.schedmd.com/squeue.html#SECTION_JOB-STATE-CODES
+        pending_states = {
+            'CONFIGURING',
+            'PENDING'
+        }
+        if state:
+            return any(s in pending_states for s in state.split(','))
+
+        return False

src/compute/compute.py

@@ -269,8 +269,11 @@ def get_job_files(headers, system_name, system_addr, job_info, output=False, use
     control_info["job_file"] = control_dict.get("Command", "command-not-found")
     control_info["job_data_out"] = ""
     control_info["job_data_err"] = ""
+    state = control_dict.get("JobState", "UNKNOWN")
 
-    if output:
+    # Return empty output if the job is pending, since the files may belong
+    # to an older job
+    if output and not scheduler.job_is_pending(state):
         # to add data from StdOut and StdErr files in Task
         # this is done when GET compute/jobs is triggered.
         #
@@ -741,7 +744,7 @@ def list_job_task(headers,system_name, system_addr,action,task_id,pageSize,pageN
     jobs = {}
     for job_index, jobinfo in enumerate(jobList):
         # now looking for log and err files location
-        jobinfo = get_job_files(headers, system_name, system_addr, jobinfo, True)
+        jobinfo = get_job_files(headers, system_name, system_addr, jobinfo, output=True)
 
         # add jobinfo to the array
         jobs[str(job_index)]=jobinfo