From 883f385ff87d248fd1ede52fa2aec6a2e696d1e6 Mon Sep 17 00:00:00 2001
From: Mark Stockton <mark@cern.ch>
Date: Tue, 5 Nov 2019 18:13:49 +0100
Subject: [PATCH] Add log file checking to find failed child jobs and set
mother return code to this result Tested by killing by force a child
process
---
.../TrigTransform/python/trigRecoExe.py | 25 ++++++++++++++++---
1 file changed, 22 insertions(+), 3 deletions(-)
diff --git a/HLT/Trigger/TrigTransforms/TrigTransform/python/trigRecoExe.py b/HLT/Trigger/TrigTransforms/TrigTransform/python/trigRecoExe.py
index 62e9791b844..90060002dda 100644
--- a/HLT/Trigger/TrigTransforms/TrigTransform/python/trigRecoExe.py
+++ b/HLT/Trigger/TrigTransforms/TrigTransform/python/trigRecoExe.py
@@ -17,7 +17,7 @@ import subprocess
from PyJobTransforms.trfExe import athenaExecutor
#imports for preExecute
-from PyJobTransforms.trfUtils import asetupReport, cvmfsDBReleaseCheck
+from PyJobTransforms.trfUtils import asetupReport, cvmfsDBReleaseCheck, lineByLine
import PyJobTransforms.trfEnv as trfEnv
import PyJobTransforms.trfExceptions as trfExceptions
from PyJobTransforms.trfExitCodes import trfExit as trfExit
@@ -195,8 +195,27 @@ class trigRecoExecutor(athenaExecutor):
def postExecute(self):
- #TODO
- #need to check for HLTMPPU.*Child Issue in the log file and throw an error message if there so we catch that the child died
+ #Adding check for HLTMPPU.*Child Issue in the log file
+ # Throws an error message if there so we catch that the child died
+ # Also sets the return code of the mother process to mark the job as failed
+ # Is based on trfValidation.scanLogFile
+ log = self._logFileName
+ msg.debug('Now scanning logfile {0}'.format(log))
+ # Using the generator so that lines can be grabbed by subroutines if needed for more reporting
+ try:
+ myGen = lineByLine(log, substepName=self._substep)
+ except IOError as e:
+ msg.error('Failed to open transform logfile {0}: {1:s}'.format(log, e))
+ for line, lineCounter in myGen:
+ # Check to see if any of the hlt children had an issue
+ if 'Child Issue' in line > -1:
+ try:
+ signal = int((re.search('signal ([0-9]*)', line)).group(1))
+ except AttributeError:
+ #text signal not found so just return 0
+ signal = 0
+ msg.error('Detected issue with HLTChild, setting mother return code to %s' % (signal) )
+ self._rc = signal
msg.info("Check for expert-monitoring.root file")
#the BS-BS step generates the files:
--
GitLab