def rejector(url, specific, options=None):
up = componentInfo()
if specific.startswith('/'):
pass
else:
wfo = session.query(Workflow).filter(Workflow.name == specific).first()
if not wfo:
print "cannot reject",spec
return
results=[]
wfi = workflowInfo(url, wfo.name)
reqMgrClient.invalidateWorkflow(url, wfo.name, current_status=wfi.request['RequestStatus'])
#if wfi.request['RequestStatus'] in ['assignment-approved','new','completed']:
# #results.append( reqMgrClient.rejectWorkflow(url, wfo.name))
# reqMgrClient.rejectWorkflow(url, wfo.name)
#else:
# #results.append( reqMgrClient.abortWorkflow(url, wfo.name))
# reqMgrClient.abortWorkflow(url, wfo.name)
datasets = wfi.request['OutputDatasets']
for dataset in datasets:
if options.keep:
print "keeping",dataset,"in its current status"
else:
results.append( setDatasetStatus(dataset, 'INVALID') )
if all(map(lambda result : result in ['None',None,True],results)):
wfo.status = 'forget'
session.commit()
print wfo.name,"and",datasets,"are rejected"
if options and options.clone:
schema = wfi.getSchema()
schema['Requestor'] = os.getenv('USER')
schema['Group'] = 'DATAOPS'
schema['OriginalRequestName'] = wfo.name
if 'ProcessingVersion' in schema:
schema['ProcessingVersion']+=1
else:
schema['ProcessingVersion']=2
##schema.pop('RequestDate') ## ok then, let's not reset the time stamp
if options.Memory:
schema['Memory'] = options.Memory
response = reqMgrClient.submitWorkflow(url, schema)
m = re.search("details\/(.*)\'",response)
if not m:
print "error in cloning",wfo.name
print response
return
newWorkflow = m.group(1)
data = reqMgrClient.setWorkflowApproved(url, newWorkflow)
print data
wfo.status = 'trouble'
session.commit()
else:
print "error in rejecting",wfo.name,results
def main():
usage = "\n python %prog [options] [WORKFLOW_NAME]\n" \
"WORKFLOW_NAME: if the list file is provided this should be empty\n"
parser = OptionParser(usage=usage)
parser.add_option('-f', '--file', help='Text file of workflows to Reject and Clone', dest='file')
parser.add_option('-c', '--clone', help='Are the workflows going to be cloned? The default value is False',action="store_true", dest='clone', default=False)
parser.add_option('-i', '--invalidate', help='Invalidate datasets? The default value is False',action="store_true", dest='invalidate', default=False)
parser.add_option("-u", "--user", dest="user",help="The user for creating the clone, if empty it will use the OS user running the script")
parser.add_option("-g", "--group", dest="group", default='DATAOPS',help="The group for creating the clone, if empty it will, use 'DATAOPS' by default")
(options, args) = parser.parse_args()
# Check the arguments, get info from them
if options.file:
try:
workflows = [l.strip() for l in open(options.file) if l.strip()]
except:
parser.error("Provide a valid file of workflows")
sys.exit(1)
elif len(args) >0:
# name of workflow
workflows = [args[0]]
else:
parser.error("Provide the workflow of a file of workflows")
sys.exit(1)
if not options.user:
# get os username by default
uinfo = pwd.getpwuid(os.getuid())
user = uinfo.pw_name
else:
user = options.user
for workflow in workflows:
try:
workflowInfo = reqMgrClient.Workflow(workflow)
except:
print("The workflow name: "+ workflow+" is not valid.")
continue
# invalidates workflow
print("Invalidating the workflow: "+ workflow)
reqMgrClient.invalidateWorkflow(url,workflow,workflowInfo.status)
# invalidates datasets
if options.invalidate:
print("Invalidating datasets")
datasets = reqMgrClient.outputdatasetsWorkflow(url, workflow)
for dataset in datasets:
print(dataset)
dbs3.setDatasetStatus(dataset, 'INVALID', files=True)
# clones workflow
if options.clone:
print("Cloning workflow: "+ workflow)
cloned = resubmit.cloneWorkflow(workflow, user, options.group)
sys.exit(0);
def forceComplete(url, wfi):
familly = getWorkflowById( url, wfi.request['PrepID'] ,details=True)
for member in familly:
### if member['RequestName'] == wl['RequestName']: continue ## set himself out
if member['RequestDate'] < wfi.request['RequestDate']: continue
if member['RequestStatus'] in ['None',None]: continue
## then set force complete all members
if member['RequestStatus'] in ['running-opened','running-closed']:
#sendEmail("force completing","%s is worth force completing\n%s"%( member['RequestName'] , percent_completions))
print "setting",member['RequestName'],"force-complete"
reqMgrClient.setWorkflowForceComplete(url, member['RequestName'])
elif member['RequestStatus'] in ['acquired','assignment-approved']:
print "rejecting",member['RequestName']
reqMgrClient.invalidateWorkflow(url, member['RequestName'], current_status=member['RequestStatus'])
def forceComplete(url, wfi):
familly = getWorkflowById(url, wfi.request['PrepID'], details=True)
for member in familly:
### if member['RequestName'] == wl['RequestName']: continue ## set himself out
if member['RequestDate'] < wfi.request['RequestDate']: continue
if member['RequestStatus'] in ['None', None]: continue
## then set force complete all members
if member['RequestStatus'] in ['running-opened', 'running-closed']:
#sendEmail("force completing","%s is worth force completing\n%s"%( member['RequestName'] , percent_completions))
print "setting", member['RequestName'], "force-complete"
reqMgrClient.setWorkflowForceComplete(url, member['RequestName'])
elif member['RequestStatus'] in ['acquired', 'assignment-approved']:
print "rejecting", member['RequestName']
reqMgrClient.invalidateWorkflow(
url,
member['RequestName'],
current_status=member['RequestStatus'])
def main():
usage = "\n python %prog [options] [WORKFLOW_NAME]\n" \
"WORKFLOW_NAME: if the list file is provided this should be empty\n"
parser = OptionParser(usage=usage)
parser.add_option('-f',
'--file',
help='Text file of workflows to Reject and Clone',
dest='file')
parser.add_option(
'-c',
'--clone',
help='Are the workflows going to be cloned? The default value is False',
action="store_true",
dest='clone',
default=False)
parser.add_option('-i',
'--invalidate',
help='Invalidate datasets? The default value is False',
action="store_true",
dest='invalidate',
default=False)
parser.add_option(
"-u",
"--user",
dest="user",
help=
"The user for creating the clone, if empty it will use the OS user running the script"
)
parser.add_option(
"-g",
"--group",
dest="group",
default='DATAOPS',
help=
"The group for creating the clone, if empty it will, use 'DATAOPS' by default"
)
parser.add_option(
"-m",
"--memory",
dest="memory",
help=
"Set max memory for the clone. At assignment, this will be used to calculate maxRSS = memory*1024"
)
(options, args) = parser.parse_args()
# Check the arguments, get info from them
if options.file:
try:
workflows = [l.strip() for l in open(options.file) if l.strip()]
except:
parser.error("Provide a valid file of workflows")
sys.exit(1)
elif len(args) > 0:
# name of workflow
workflows = [args[0]]
else:
parser.error("Provide the workflow of a file of workflows")
sys.exit(1)
if not options.user:
# get os username by default
uinfo = pwd.getpwuid(os.getuid())
user = uinfo.pw_name
else:
user = options.user
for workflow in workflows:
try:
workflowInfo = reqMgrClient.Workflow(workflow)
except:
print("The workflow name: " + workflow + " is not valid.")
continue
# invalidates workflow
print("Invalidating the workflow: " + workflow)
reqMgrClient.invalidateWorkflow(url, workflow, workflowInfo.status)
# invalidates datasets
if options.invalidate:
print("Invalidating datasets")
datasets = reqMgrClient.outputdatasetsWorkflow(url, workflow)
for dataset in datasets:
print(dataset)
dbs3.setDatasetStatus(dataset, 'INVALID', files=True)
# clones workflow
if options.clone:
print("Cloning workflow: " + workflow)
if options.memory:
mem = options.memory
else:
mem = workflowInfo.info["Memory"]
cloned = resubmit.cloneWorkflow(workflow,
user,
options.group,
memory=mem)
sys.exit(0)
def invalidator(url, invalid_status='INVALID'):
use_mcm = True
up = componentInfo(mcm=use_mcm)
if not up.check(): return
mcm = McMClient(dev=False)
invalids = mcm.getA('invalidations',query='status=announced')
print len(invalids),"Object to be invalidated"
text_to_batch = defaultdict(str)
text_to_request = defaultdict(str)
for invalid in invalids:
acknowledge= False
pid = invalid['prepid']
batch_lookup = invalid['prepid']
text = ""
if invalid['type'] == 'request':
wfn = invalid['object']
print "need to invalidate the workflow",wfn
wfo = session.query(Workflow).filter(Workflow.name == wfn).first()
if wfo:
## set forget of that thing (although checkor will recover from it)
print "setting the status of",wfo.status,"to forget"
wfo.status = 'forget'
session.commit()
else:
## do not go on like this, do not acknoledge it
print wfn,"is set to be rejected, but we do not know about it yet"
#continue
wfi = workflowInfo(url, wfn)
success = "not rejected"
## to do, we should find a way to reject the workflow and any related acdc
success = reqMgrClient.invalidateWorkflow(url, wfn, current_status = wfi.request['RequestStatus'])
## need to find the whole familly and reject the whole gang
familly = getWorkflowById( url, wfi.request['PrepID'] , details=True)
for fwl in familly:
## take out all acdc
if fwl['RequestDate'] < wfi.request['RequestDate']:continue
if fwl['RequestType']!='Resubmission': continue
print "rejecting",fwl['RequestName']
success = reqMgrClient.invalidateWorkflow(url, fwl['RequestName'], current_status=fwl['RequestStatus'])
print success
wfi.sendLog('invalidator',"rejection is performed from McM invalidations request")
acknowledge= True
text = "The workflow %s (%s) was rejected due to invalidation in McM" % ( wfn, pid )
batch_lookup = wfn ##so that the batch id is taken as the one containing the workflow name
elif invalid['type'] == 'dataset':
dataset = invalid['object']
if '?' in dataset: continue
if 'None' in dataset: continue
if 'None-' in dataset: continue
if 'FAKE-' in dataset: continue
print "setting",dataset,"to",invalid_status
success = setDatasetStatus(dataset , invalid_status )
if success:
acknowledge= True
text = "The dataset %s (%s) was set INVALID due to invalidation in McM" % ( dataset, pid )
else:
print "invalidation of",dataset,"did not go so well"
else:
print "\t\t",invalid['type']," type not recognized"
if acknowledge:
## acknoldge invalidation in mcm, provided we can have the api
print "acknowledgment to mcm"
mcm.get('/restapi/invalidations/acknowledge/%s'%( invalid['_id'] ))
# prepare the text for batches
batches = []
batches.extend(mcm.getA('batches',query='contains=%s'%batch_lookup))
batches = filter(lambda b : b['status'] in ['announced','done','reset'], batches)
if len(batches):
bid = batches[-1]['prepid']
print "batch nofication to",bid
text_to_batch[bid] += text+"\n\n"
# prepare the text for requests
text_to_request[pid] += text+"\n\n"
for bid,text in text_to_batch.items():
if not text: continue
text += '\n This is an automated message'
mcm.put('/restapi/batches/notify',{ "notes" : text, "prepid" : bid})
pass
for pid,text in text_to_request.items():
if not text: continue
text += '\n This is an automated message'
mcm.put('/restapi/requests/notify',{ "message" : text, "prepids" : [pid]})
def rejector(url, specific, options=None):
#use_mcm = True
#up = componentInfo(mcm=use_mcm, soft=['mcm'])
up = componentInfo()
if not up.check(): return
#use_mcm = up.status['mcm']
#mcm = McMClient(dev=False) if use_mcm else None
if specific and specific.startswith('/'):
## this is for a dataset
print setDatasetStatus(specific, 'INVALID')
return
if options.filelist:
wfs = []
for line in filter(None, open(options.filelist).read().split('\n')):
print line
wfs.extend( session.query(Workflow).filter(Workflow.name.contains(line)).all())
elif specific:
wfs = session.query(Workflow).filter(Workflow.name.contains(specific)).all()
else:
wfs = session.query(Workflow).filter(Workflow.status == 'assistance-clone').all()
#wfs.extend( session.query(Workflow).filter(Workflow.status == 'assistance-reject').all())
## be careful then on clone case by case
options.clone = True
print "not supposed to function yet"
return
print len(wfs),"to reject"
if len(wfs)>1:
print "\n".join( [wfo.name for wfo in wfs] )
answer = raw_input('Reject these')
if not answer.lower() in ['y','yes']:
return
for wfo in wfs:
#wfo = session.query(Workflow).filter(Workflow.name == specific).first()
if not wfo:
print "cannot reject",spec
return
results=[]
wfi = workflowInfo(url, wfo.name)
datasets = set(wfi.request['OutputDatasets'])
reqMgrClient.invalidateWorkflow(url, wfo.name, current_status=wfi.request['RequestStatus'])
comment=""
if options.comments: comment = ", reason: "+options.comments
wfi.sendLog('rejector','invalidating the workflow by unified operator%s'%comment)
## need to find the whole familly and reject the whole gang
familly = getWorkflowById( url, wfi.request['PrepID'] , details=True)
for fwl in familly:
if fwl['RequestDate'] < wfi.request['RequestDate']: continue
if fwl['RequestType']!='Resubmission': continue
## does not work on second order acd
if 'OriginalRequestName' in fwl and fwl['OriginalRequestName'] != wfi.request['RequestName']: continue
print "rejecting",fwl['RequestName']
reqMgrClient.invalidateWorkflow(url, fwl['RequestName'], current_status=fwl['RequestStatus'], cascade=False)
datasets.update( fwl['OutputDatasets'] )
for dataset in datasets:
if options.keep:
print "keeping",dataset,"in its current status"
else:
results.append( setDatasetStatus(dataset, 'INVALID') )
pass
if all(map(lambda result : result in ['None',None,True],results)):
print wfo.name,"and",datasets,"are rejected"
if options and options.clone:
wfo.status = 'trouble'
session.commit()
schema = wfi.getSchema()
schema['Requestor'] = os.getenv('USER')
schema['Group'] = 'DATAOPS'
schema['OriginalRequestName'] = wfo.name
if 'ProcessingVersion' in schema:
schema['ProcessingVersion'] = int(schema['ProcessingVersion'])+1 ## dubious str->int conversion
else:
schema['ProcessingVersion']=2
for k in schema.keys():
if k.startswith('Team'):
schema.pop(k)
if k.startswith('checkbox'):
schema.pop(k)
## a few tampering of the original request
if options.Memory:
if schema['RequestType'] == 'TaskChain':
it=1
while True:
t = 'Task%d'%it
it+=1
if t in schema:
schema[t]['Memory'] = options.Memory
else:
break
else:
schema['Memory'] = options.Memory
if options.Multicore:
## to do : set it properly in taslchains
schema['Multicore'] = options.Multicore
if options.deterministic:
if schema['RequestType'] == 'TaskChain':
schema['Task1']['DeterministicPileup'] = True
if options.EventsPerJob:
if schema['RequestType'] == 'TaskChain':
schema['Task1']['EventsPerJob'] = options.EventsPerJob
else:
schema['EventsPerJob'] = options.EventsPerJob
if options.EventAwareLumiBased:
schema['SplittingAlgo'] = 'EventAwareLumiBased'
if options.TimePerEvent:
schema['TimePerEvent'] = options.TimePerEvent
if options.ProcessingString:
schema['ProcessingString'] = options.ProcessingString
if options.AcquisitionEra:
schema['AcquisitionEra'] = options.AcquisitionEra
if options.runs:
schema['RunWhitelist'] = map(int,options.runs.split(','))
if options.PrepID:
schema['PrepID'] =options.PrepID
if schema['RequestType'] == 'TaskChain' and options.no_output:
ntask = schema['TaskChain']
for it in range(1,ntask-1):
schema['Task%d'%it]['KeepOutput'] = False
schema['TaskChain'] = ntask-1
schema.pop('Task%d'%ntask)
## update to the current priority
schema['RequestPriority'] = wfi.request['RequestPriority']
## drop shit on the way to reqmgr2
paramBlacklist = ['BlockCloseMaxEvents', 'BlockCloseMaxFiles', 'BlockCloseMaxSize', 'BlockCloseMaxWaitTime',
'CouchWorkloadDBName', 'CustodialGroup', 'CustodialSubType', 'Dashboard',
'GracePeriod', 'HardTimeout', 'InitialPriority', 'inputMode', 'MaxMergeEvents', 'MaxMergeSize',
'MaxRSS', 'MaxVSize', 'MinMergeSize', 'NonCustodialGroup', 'NonCustodialSubType',
'OutputDatasets', 'ReqMgr2Only', 'RequestDate' 'RequestorDN', 'RequestName', 'RequestStatus',
'RequestTransition', 'RequestWorkflow', 'SiteWhitelist', 'SoftTimeout', 'SoftwareVersions',
'SubscriptionPriority', 'Team', 'timeStamp', 'TrustSitelists', 'TrustPUSitelists',
'TotalEstimatedJobs', 'TotalInputEvents', 'TotalInputLumis', 'TotalInputFiles']
for p in paramBlacklist:
if p in schema:
schema.pop( p )
#pass
print "submitting"
if (options.to_stepchain and (schema['RequestType']=='TaskChain')):
## transform the schema into StepChain schema
print "Transforming a TaskChain into a StepChain"
schema['RequestType'] = 'StepChain'
schema['StepChain'] = schema.pop('TaskChain')
step=1
while True:
if 'Task%d'%step in schema:
schema['Step%d'%step] = schema.pop('Task%d'%step)
schema['Step%d'%step]['StepName'] = schema['Step%d'%step].pop('TaskName')
if 'InputTask' in schema['Step%d'%step]:
schema['Step%d'%step]['InputStep'] = schema['Step%d'%step].pop('InputTask')
if not 'KeepOutput' in schema['Step%d'%step]:
## this is a weird translation capability. Absence of keepoutput in step means : keep the output. while in TaskChain absence means : drop
schema['Step%d'%step]['KeepOutput'] = False
step+=1
else:
break
print json.dumps( schema, indent=2 )
newWorkflow = reqMgrClient.submitWorkflow(url, schema)
if not newWorkflow:
print "error in cloning",wfo.name
print json.dumps( schema, indent=2 )
return
print newWorkflow
data = reqMgrClient.setWorkflowApproved(url, newWorkflow)
print data
wfi.sendLog('rejector','Cloned into %s by unified operator %s'%( newWorkflow, comment ))
wfi.notifyRequestor('Cloned into %s by unified operator %s'%( newWorkflow, comment ),do_batch=False)
else:
wfo.status = 'forget'
wfi.notifyRequestor('Rejected by unified operator %s'%( comment ),do_batch=False)
session.commit()
else:
print "error in rejecting",wfo.name,results
def actor(url,options=None):
if userLock('actor'): return
up = componentInfo(mcm=False, soft=['mcm'])
if not up.check(): return
# CI = campaignInfo()
SI = siteInfo()
UC = unifiedConfiguration()
# Need to look at the actions page https://vocms0113.cern.ch:80/getaction (can add ?days=20) and perform any actions listed
try:
action_list = json.loads(os.popen('curl -s -k https://vocms0113.cern.ch:80/getaction?days=15').read())
## now we have a list of things that we can take action on
except:
print "Not able to load action list :("
sendLog('actor','Not able to load action list', level='critical')
return
print action_list
if not action_list:
print "EMPTY!"
return
for wfname in action_list:
print '-'*100
print "Looking at",wfname,"for recovery options"
to_clone = False
to_acdc = False
for key in action_list[wfname]:
if key == 'Parameters':
tasks = action_list[wfname][key]
elif key == 'Action' and action_list[wfname][key] == 'acdc':
print "Going to create ACDCs for ", wfname
to_acdc = True
elif key == 'Action' and action_list[wfname][key] == 'clone':
print "Going to clone ", wfname
to_clone = True
if not to_acdc and not to_clone:
sendLog('actor','Action submitted for something other than acdc and clone for workflow %s'%wfname,level='critical')
print "Can only do acdcs and clones! Skipping workflow ",wfname
continue
if not tasks:
sendLog('actor','Empty action submitted for workflow %s'%wfname,level='critical')
print "Moving on. Parameters is blank for " + wfname
continue
wfi = workflowInfo(url, wfname)
recover = True
message_to_ops = ""
message_to_user = ""
#===========================================================
if to_clone and options.do:
print "Let's try kill and clone: "
wfi.sendLog('actor','Going to clone %s'%wfname)
results=[]
datasets = set(wfi.request['OutputDatasets'])
comment=""
if 'comment' in tasks: comment = ", reason: "+ tasks['comment']
wfi.sendLog('actor',"invalidating the workflow by traffic controller %s"%comment)
#Reject all workflows in the family
#first reject the original workflow.
reqMgrClient.invalidateWorkflow(url, wfi.request['RequestName'], current_status=wfi.request['RequestStatus'], cascade=False)
#Then reject any ACDCs associated with that workflow
if 'ACDCs' in action_list[wfname]:
children = action_list[wfname]['ACDCs']
for child in children:
wfi.sendLog('actor',"rejecting %s"%child)
wfi_acdc = workflowInfo(url, child)
reqMgrClient.invalidateWorkflow(url, wfi_acdc.request['RequestName'], current_status=wfi_acdc.request['RequestStatus'], cascade=False)
datasets.update( wfi_acdc.request['OutputDatasets'] )
#Invalidate all associated output datasets
for dataset in datasets:
results.append( setDatasetStatus(dataset, 'INVALID') )
if all(map(lambda result : result in ['None',None,True],results)):
wfi.sendLog('actor',"%s and children are rejected"%wfname)
cloned = None
try:
cloned = singleClone(url, wfname, tasks, comment, options.do)
except:
sendLog('actor','Failed to create clone for %s! Check logs for more information. Action will need to be resubmitted.'%wfname,level='critical')
wfi.sendLog('actor','Failed to create clone for %s!'%wfname)
remove_action(wfname)
if not cloned:
recover = False
wfi.sendLog('actor','Failed to create clone for %s!'%wfname)
sendLog('actor','Failed to create clone for %s!'%wfname,level='critical')
else:
wfi.sendLog('actor',"Workflow %s cloned"%wfname)
#===========================================================
elif to_acdc:
if 'AllSteps' in tasks:
allTasksDefaults = tasks['AllSteps']
tasks.pop('AllSteps')
for setting in allTasksDefaults:
for task in tasks:
if setting in tasks[task]:
tasks[task][setting] = allTasksDefaults[setting]
else:
tasks[task].append({setting:allTasksDefaults[setting]})
print "Tasks is "
print tasks
all_tasks = wfi.getAllTasks()
## need a way to verify that this is the first round of ACDC, since the second round will have to be on the ACDC themselves
try:
WMErr = wfi.getWMErrors()
# print WMErr
except:
sendLog('actor','Cannot create ACDCS for %s because WMErr cannot be reached.'%wfname,level='critical')
continue
if not WMErr:
sendLog('actor','Cannot create ACDCS for %s because WMErr is blank.'%wfname,level='critical')
print "Moving on. WMErr is blank"
continue
try:
where_to_run, missing_to_run,missing_to_run_at = wfi.getRecoveryInfo()
print "Where to run = "
print where_to_run
except:
sendLog('actor','Cannot create ACDCS for %s because recovery info cannot be found.'%wfname,level='critical')
print "Moving on. Cannot access recovery info for " + wfname
continue
if not where_to_run:
sendLog('actor','Cannot create ACDCS for %s because site list cannot be found.'%wfname,level='critical')
print "Moving on. where to run is blank"
continue
message_to_ops = ""
message_to_user = ""
num_tasks_to_recover = 0
for task in WMErr:
if 'LogCollect' in task: continue
if 'Cleanup' in task: continue
if not 'jobfailed' in WMErr[task]:
continue
else:
num_tasks_to_recover += 1
# print "Task to recover: " + task
if not num_tasks_to_recover:
print "\tno error for",wfname
# recover = False
if 'LheInputFiles' in wfi.request and wfi.request['LheInputFiles']:
## we do not try to recover pLHE
sendLog('actor','Cannot create ACDCS for %s because it is a pLHE workflow.'%wfname,level='critical')
print "We don't try to recover pLHE. Moving on."
recover = False
# sendEmail('cannot submit action', '%s is a pLHE workflow. We do not try to recover pLHE'%wfname)
# if wfi.request['RequestType'] in ['ReReco']:
# recover= False
# print 'cannot submit action. ReReco'
# sendEmail('cannot submit action', '%s is request type ReReco'%wfname)
recovering = set()
for task in tasks:
assign_to_sites = set()
print "Task names is " + task
fulltaskname = '/' + wfname + '/' + task
# print "Full task name is " + fulltaskname
wrong_task = False
for task_info in all_tasks:
if fulltaskname == task_info.pathName:
if task_info.taskType not in ['Processing','Production','Merge']:
wrong_task=True
wfi.sendLog('actor', "Skipping task %s because the taskType is %s. Can only ACDC Processing, Production, or Merge tasks"%( fulltaskname, task_info.taskType))
if wrong_task:
continue
print tasks[task]
actions = tasks[task]
for action in actions:
if action.startswith('sites'):
if type(actions[action]) != list:
assign_to_sites=[SI.SE_to_CE(actions[action])]
else:
assign_to_sites=list(set([SI.SE_to_CE(site) for site in actions[action]]))
# if action.startswith('mem') and actions[action] != "" and actions[action] != 'Same' and wfi.request['RequestType'] in ['TaskChain']:
# recover = False;
# print "Skipping %s for now until Allie fixes memory parameter for TaskChain ACDCs."%wfname
# wfi.sendLog('actor',"Skipping %s for now until Allie fixes memory parameter for TaskChain ACDCs."%wfname)
if not 'sites' in actions:
assign_to_sites = list(set([SI.SE_to_CE(site) for site in where_to_run[task]]))
print "Found",sorted(assign_to_sites),"as sites where to run the ACDC at, from the acdc doc of ",wfname
print "Going to run at",sorted(assign_to_sites)
if recover:
print "Initiating recovery"
acdc = singleRecovery(url, fulltaskname, wfi.request, actions, do = options.do)
if not acdc:
if options.do:
if recovering:
print wfname + " has been partially ACDC'ed. Needs manual attention."
sendLog('actor', "%s has had %s/%s recoveries %s only"%( wfname, len(recovering), num_tasks_to_recover, list(recovering)), level='critical')
wfi.sendLog('actor', "%s has had %s/%s recoveries %s only"%( wfname, len(recovering), num_tasks_to_recover, list(recovering)))
break
else:
print wfname + " failed recovery once"
recover = False
break
else:
print "no action to take further"
# sendLog('recoveror', "ACDC for %s can be done automatically"% wfname, level='critical')
continue
else: #ACDC was made correctly. Now we have to assign it.
wfi.sendLog('actor','ACDC created for task %s. Actions taken \n%s'%(fulltaskname,list(actions)))
team = wfi.request['Teams'][0]
parameters={
'SiteWhitelist' : sorted(assign_to_sites),
'AcquisitionEra' : wfi.acquisitionEra(),
'ProcessingString' : wfi.processingString(),
'MergedLFNBase' : wfi.request['MergedLFNBase'],
'ProcessingVersion' : wfi.request['ProcessingVersion'],
}
## hackery for ACDC merge assignment
if wfi.request['RequestType'] == 'TaskChain' and 'Merge' in task.split('/')[-1]:
parameters['AcquisitionEra'] = None
parameters['ProcessingString'] = None
## xrootd setttings on primary and secondary
if 'xrootd' in actions:
if actions['xrootd'] == 'enabled':
print "Going to assign via xrootd"
parameters['TrustSitelists'] = True
elif actions['xrootd'] == 'disabled':
parameters['TrustSitelists'] = False
elif ('TrustSitelists' in wfi.request and wfi.request['TrustSitelists']=='true'):
parameters['TrustSitelists'] = True
else:
parameters['TrustSitelists'] = False
if 'TrustPUSitelists' in wfi.request and wfi.request['TrustPUSitelists']:
parameters['TrustPUSitelists'] = True
if options.ass:
print "really doing the assignment of the ACDC",acdc
parameters['execute']=True
wfi.sendLog('actor',"%s was assigned for recovery"% acdc)
else:
print "no assignment done with this ACDC",acdc
sendLog('actor',"%s needs to be assigned"%(acdc), level='critical')
continue
# print parameters
result = reqMgrClient.assignWorkflow(url, acdc, team, parameters)
if not result:
print acdc,"was not assigned"
sendLog('actor',"%s needs to be assigned"%(acdc), level='critical')
else:
recovering.add( acdc )
wfi.sendLog('actor',"ACDCs created for %s"%wfname)
#===========================================================
if recover and options.do:
remove_action(wfname)
if message_to_user:
print wfname,"to be notified to user(DUMMY)",message_to_user
if message_to_ops:
print 'message'
#sendEmail( "notification in recoveror" , message_to_ops, destination=['*****@*****.**'])
# sendLog('recoveror',message_to_ops,level='warning')
return
def actor(url, options=None):
if moduleLock(wait=False, silent=True)(): return
if userLock('actor'): return
up = componentInfo(soft=['mcm'])
if not up.check(): return
# CI = campaignInfo()
SI = siteInfo()
UC = unifiedConfiguration()
WC = wtcClient()
action_list = WC.get_actions()
if action_list is None:
print "Not able to load action list"
sendLog('actor', 'Not able to load action list', level='critical')
return
if options.actions:
action_list = json.loads(open(options.actions).read())
print json.dumps(action_list, indent=2)
if not action_list:
print "EMPTY!"
return
wf_list = action_list.keys()
print json.dumps(sorted(wf_list), indent=2)
if options.spec:
wf_list = [wf for wf in wf_list if options.spec in wf]
max_per_round = UC.get('max_per_round').get('actor', None)
if max_per_round:
random.shuffle(wf_list)
wf_list = wf_list[:max_per_round]
for wfname in wf_list:
print '-' * 100
print "Looking at", wfname, "for recovery options"
to_clone = False
to_acdc = False
to_force = False
to_hold = False
something_to_do = False
tasks = action_list[wfname].get('Parameters', None)
to_acdc = action_list[wfname].get('Action', None) == 'acdc'
to_clone = action_list[wfname].get('Action', None) == 'clone'
to_force = action_list[wfname].get(
'Action', None) == 'special' and action_list[wfname].get(
'Parameters', {}).get('action', None) in ['by-pass', 'bypass']
to_hold = action_list[wfname].get(
'Action', None) == 'special' and action_list[wfname].get(
'Parameters', {}).get('action', None) in ['onhold', 'on-hold']
if not to_acdc and not to_clone and not to_force and not to_hold:
sendLog(
'actor',
'Action submitted for something other than acdc, clone, bypass or hold for workflow %s'
% wfname,
level='critical')
print json.dumps(action_list[wfname], indent=2)
continue
if not tasks and to_acdc:
sendLog('actor',
'Empty action submitted for workflow %s' % wfname,
level='critical')
print "Moving on. Parameters is blank for " + wfname
continue
wfi = workflowInfo(url, wfname)
recover = True
message_to_ops = ""
message_to_user = ""
#===========================================================
if to_clone and options.do:
print "Let's try kill and clone: "
wfi.sendLog('actor', 'Going to clone %s' % wfname)
results = []
datasets = set(wfi.request['OutputDatasets'])
comment = ""
if 'comment' in tasks: comment = ", reason: " + tasks['comment']
wfi.sendLog(
'actor',
"invalidating the workflow by traffic controller %s" % comment)
#Reject all workflows in the family
#first reject the original workflow.
reqMgrClient.invalidateWorkflow(
url,
wfi.request['RequestName'],
current_status=wfi.request['RequestStatus'],
cascade=False)
#Then reject any ACDCs associated with that workflow
family = getWorkflowById(url, wfi.request['PrepID'], details=True)
for fwl in family:
print "rejecting", fwl['RequestName'], fwl['RequestStatus']
wfi.sendLog(
'actor', "rejecting %s, previous status %s" %
(fwl['RequestName'], fwl['RequestStatus']))
reqMgrClient.invalidateWorkflow(
url,
fwl['RequestName'],
current_status=fwl['RequestStatus'],
cascade=False)
datasets.update(fwl['OutputDatasets'])
#Invalidate all associated output datasets
for dataset in datasets:
results.append(setDatasetStatus(dataset, 'INVALID'))
if all(map(lambda result: result in ['None', None, True],
results)):
wfi.sendLog('actor', "%s and children are rejected" % wfname)
cloned = None
try:
cloned = singleClone(url, wfname, tasks, comment, options.do)
except Exception as e:
sendLog(
'actor',
'Failed to create clone for %s! Check logs for more information. Action will need to be resubmitted.'
% wfname,
level='critical')
wfi.sendLog('actor', 'Failed to create clone for %s!' % wfname)
print str(e)
##let's not remove the action other the workflow goes to "trouble" and the WTC cannot set the action again
if not cloned:
recover = False
wfi.sendLog('actor', 'Failed to create clone for %s!' % wfname)
sendLog('actor',
'Failed to create clone for %s!' % wfname,
level='critical')
else:
wfi.sendLog('actor', "Workflow %s cloned" % wfname)
#===========================================================
elif to_force:
wfi.sendLog('actor',
'Bypassing from workflow traffic controler request')
forcing = json.loads(
open(
'/afs/cern.ch/user/v/vlimant/public/ops/forcecomplete.json'
).read())
forcing.append(wfname)
open('/afs/cern.ch/user/v/vlimant/public/ops/forcecomplete.json',
'w').write(json.dumps(sorted(set(forcing))))
elif to_hold:
wfi.sendLog('actor',
'Holding on workflow traffic controler request')
holding = json.loads(
open('/afs/cern.ch/user/v/vlimant/public/ops/onhold.json').
read())
holding.append(wfname)
open('/afs/cern.ch/user/v/vlimant/public/ops/onhold.json',
'w').write(json.dumps(sorted(set(holding))))
#===========================================================
elif to_acdc:
if 'AllSteps' in tasks:
allTasksDefaults = tasks['AllSteps']
tasks.pop('AllSteps')
for setting in allTasksDefaults:
for task in tasks:
if setting in tasks[task]:
tasks[task][setting] = allTasksDefaults[setting]
else:
tasks[task].append(
{setting: allTasksDefaults[setting]})
print "Tasks is "
print json.dumps(tasks, indent=2)
all_tasks = wfi.getAllTasks()
## need a way to verify that this is the first round of ACDC, since the second round will have to be on the ACDC themselves
try:
WMErr = wfi.getWMErrors()
# print WMErr
except:
sendLog(
'actor',
'Cannot create ACDCS for %s because WMErr cannot be reached.'
% wfname,
level='critical')
continue
if not WMErr:
wfi.sendLog('actor', 'WMErrors is blank for %s.' % wfname)
print "FYI getWMErrors is blank. Presumably there are only unreported errors"
# continue
try:
where_to_run, missing_to_run, missing_to_run_at = wfi.getRecoveryInfo(
)
print "Where to run = "
print where_to_run
except:
sendLog(
'actor',
'Cannot create ACDCS for %s because recovery info cannot be found.'
% wfname,
level='critical')
print "Moving on. Cannot access recovery info for " + wfname
continue
if not where_to_run:
sendLog(
'actor',
'Cannot create ACDCS for %s because site list cannot be found.'
% wfname,
level='critical')
print "Moving on. where to run is blank"
continue
message_to_ops = ""
message_to_user = ""
num_tasks_to_recover = 0
if WMErr:
for task in WMErr:
if 'LogCollect' in task: continue
if 'Cleanup' in task: continue
if not 'jobfailed' in WMErr[task]:
continue
else:
num_tasks_to_recover += 1
# print "Task to recover: " + task
if not num_tasks_to_recover:
print "\tno error for", wfname
# recover = False
if 'LheInputFiles' in wfi.request and wfi.request['LheInputFiles']:
## we do not try to recover pLHE
sendLog(
'actor',
'Cannot create ACDCS for %s because it is a pLHE workflow.'
% wfname,
level='critical')
print "We don't try to recover pLHE. Moving on."
recover = False
# sendEmail('cannot submit action', '%s is a pLHE workflow. We do not try to recover pLHE'%wfname)
# if wfi.request['RequestType'] in ['ReReco']:
# recover= False
# print 'cannot submit action. ReReco'
# sendEmail('cannot submit action', '%s is request type ReReco'%wfname)
recovering = set()
for task in tasks:
assign_to_sites = set()
print "Task names is " + task
fulltaskname = '/' + wfname + '/' + task
# print "Full task name is " + fulltaskname
wrong_task = False
for task_info in all_tasks:
if fulltaskname == task_info.pathName:
if task_info.taskType not in [
'Processing', 'Production', 'Merge'
]:
wrong_task = True
wfi.sendLog(
'actor',
"Skipping task %s because the taskType is %s. Can only ACDC Processing, Production, or Merge tasks"
% (fulltaskname, task_info.taskType))
if wrong_task:
continue
print tasks[task]
actions = tasks[task]
for action in actions:
if action.startswith('sites'):
if type(actions[action]) != list:
assign_to_sites = [SI.SE_to_CE(actions[action])]
else:
assign_to_sites = list(
set([
SI.SE_to_CE(site)
for site in actions[action]
]))
# if action.startswith('mem') and actions[action] != "" and actions[action] != 'Same' and wfi.request['RequestType'] in ['TaskChain']:
# recover = False;
# print "Skipping %s for now until Allie fixes memory parameter for TaskChain ACDCs."%wfname
# wfi.sendLog('actor',"Skipping %s for now until Allie fixes memory parameter for TaskChain ACDCs."%wfname)
if not 'sites' in actions:
assign_to_sites = list(
set([SI.SE_to_CE(site)
for site in where_to_run[task]]))
print "Found", sorted(
assign_to_sites
), "as sites where to run the ACDC at, from the acdc doc of ", wfname
print "Going to run at", sorted(assign_to_sites)
if recover:
print "Initiating recovery"
acdc = singleRecovery(url,
fulltaskname,
wfi.request,
actions,
do=options.do)
if not acdc:
if options.do:
if recovering:
print wfname + " has been partially ACDC'ed. Needs manual attention."
sendLog(
'actor',
"%s has had %s/%s recoveries %s only" %
(wfname, len(recovering),
num_tasks_to_recover, list(recovering)),
level='critical')
wfi.sendLog(
'actor',
"%s has had %s/%s recoveries %s only" %
(wfname, len(recovering),
num_tasks_to_recover, list(recovering)))
break
else:
print wfname + " failed recovery once"
recover = False
break
else:
print "no action to take further"
# sendLog('recoveror', "ACDC for %s can be done automatically"% wfname, level='critical')
continue
else: #ACDC was made correctly. Now we have to assign it.
wfi.sendLog(
'actor',
'ACDC created for task %s. Actions taken \n%s' %
(fulltaskname, json.dumps(actions)))
#team = wfi.request['Teams'][0]
team = 'production'
parameters = {
'SiteWhitelist': sorted(assign_to_sites),
'AcquisitionEra': wfi.acquisitionEra(),
'ProcessingString': wfi.processingString(),
'MergedLFNBase': wfi.request['MergedLFNBase'],
'ProcessingVersion':
wfi.request['ProcessingVersion'],
}
## hackery for ACDC merge assignment
if wfi.request[
'RequestType'] == 'TaskChain' and 'Merge' in task.split(
'/')[-1]:
parameters['AcquisitionEra'] = None
parameters['ProcessingString'] = None
## xrootd setttings on primary and secondary
if 'xrootd' in actions:
if actions['xrootd'] == 'enabled':
print "Going to assign via xrootd"
parameters['TrustSitelists'] = True
elif actions['xrootd'] == 'disabled':
parameters['TrustSitelists'] = False
elif ('TrustSitelists' in wfi.request
and wfi.request['TrustSitelists'] == 'true'):
parameters['TrustSitelists'] = True
else:
parameters['TrustSitelists'] = False
if 'secondary' in actions:
if actions['secondary'] == 'enabled':
print 'Enabling reading the secondary input via xrootd'
parameters['TrustPUSitelists'] = True
elif actions['secondary'] == 'disabled':
parameters['TrustPUSitelists'] = False
#in case secondary is blank or not set to enabled or disabled
elif 'TrustPUSitelists' in wfi.request and wfi.request[
'TrustPUSitelists']:
parameters['TrustPUSitelists'] = True
elif 'TrustPUSitelists' in wfi.request and wfi.request[
'TrustPUSitelists']:
parameters['TrustPUSitelists'] = True
if options.ass:
print "really doing the assignment of the ACDC", acdc
parameters['execute'] = True
wfi.sendLog('actor',
"%s was assigned for recovery" % acdc)
else:
print "no assignment done with this ACDC", acdc
sendLog('actor',
"%s needs to be assigned" % (acdc),
level='critical')
continue
# print parameters
result = reqMgrClient.assignWorkflow(
url, acdc, team, parameters)
if not result:
print acdc, "was not assigned"
sendLog('actor',
"%s needs to be assigned" % (acdc),
level='critical')
else:
recovering.add(acdc)
wfi.sendLog('actor', "ACDCs created for %s" % wfname)
#===========================================================
if recover and options.do:
r = WC.remove_action(wfname)
if not r:
sendLog(
'actor',
'not able to remove the action, interlocking the module',
level='critical')
os.system('touch %s/actor.failed-%s.lock' %
(base_eos_dir, os.getpid()))
sys.exit(-1)
if message_to_user:
print wfname, "to be notified to user(DUMMY)", message_to_user
if message_to_ops:
print 'message'
#sendEmail( "notification in recoveror" , message_to_ops, destination=['*****@*****.**'])
# sendLog('recoveror',message_to_ops,level='warning')
return
logs= filter(lambda change : change["Status"]==status, wf['RequestTransition'])
if logs:
return logs[-1]['UpdateTime']
else:
return None
delays={'assignment-approved' : (7,14),
'new':(7,14),
'completed':(14,21),
'closed-out':(14,21),
}
warnings=defaultdict(set)
for checkin,(warn,timeout) in delays.items():
wfs = getWorkflows(url, checkin, user=None, details=True)
for wf in wfs:
if not 'backfill' in wf['RequestName'].lower(): continue
transition = transitiontime(wf,checkin)
if transition and (now - transition)>(timeout*24*60*60):
## that can go away
print wf['RequestName'],"is old enough to be removed",wf['RequestStatus']
reqMgrClient.invalidateWorkflow(url, wf['RequestName'], current_status=wf['RequestStatus'])
elif transition and (now - transition)>(warn*24*60*60):
## warn requester
print wf['RequestName'],"is old enough to be removed",wf['RequestStatus']
warnings[wf['Requestor']].add( wf['RequestName'] )
for who in warnings:
sendEmail('Old Backfill in the system','The following backfill should be removed or moved to rejected/announced\n\n%s'%('\n'.join(sorted(warnings[who]))), destination=[who+'@cern.ch'])
def invalidator(url, invalid_status='INVALID'):
use_mcm = True
up = componentInfo(mcm=use_mcm)
if not up.check(): return
mcm = McMClient(dev=False)
invalids = mcm.getA('invalidations', query='status=announced')
print len(invalids), "Object to be invalidated"
text_to_batch = defaultdict(str)
text_to_request = defaultdict(str)
for invalid in invalids:
acknowledge = False
pid = invalid['prepid']
batch_lookup = invalid['prepid']
text = ""
if invalid['type'] == 'request':
wfn = invalid['object']
print "need to invalidate the workflow", wfn
wfo = session.query(Workflow).filter(Workflow.name == wfn).first()
if wfo:
## set forget of that thing (although checkor will recover from it)
print "setting the status of", wfo.status, "to forget"
wfo.status = 'forget'
session.commit()
else:
## do not go on like this, do not acknoledge it
print wfn, "is set to be rejected, but we do not know about it yet"
#continue
wfi = workflowInfo(url, wfn)
success = "not rejected"
## to do, we should find a way to reject the workflow and any related acdc
success = reqMgrClient.invalidateWorkflow(
url, wfn, current_status=wfi.request['RequestStatus'])
## need to find the whole familly and reject the whole gang
familly = getWorkflowById(url, wfi.request['PrepID'], details=True)
for fwl in familly:
if fwl['RequestDate'] < wfi.request['RequestDate']: continue
if fwl['RequestType'] != 'Resubmission': continue
print "rejecting", fwl['RequestName']
success = reqMgrClient.invalidateWorkflow(
url,
fwl['RequestName'],
current_status=fwl['RequestStatus'])
print success
acknowledge = True
text = "The workflow %s (%s) was rejected due to invalidation in McM" % (
wfn, pid)
batch_lookup = wfn ##so that the batch id is taken as the one containing the workflow name
elif invalid['type'] == 'dataset':
dataset = invalid['object']
if '?' in dataset: continue
if 'None' in dataset: continue
if 'None-' in dataset: continue
if 'FAKE-' in dataset: continue
print "setting", dataset, "to", invalid_status
success = setDatasetStatus(dataset, invalid_status)
if success:
acknowledge = True
text = "The dataset %s (%s) was set INVALID due to invalidation in McM" % (
dataset, pid)
else:
print "invalidation of", dataset, "did not go so well"
else:
print "\t\t", invalid['type'], " type not recognized"
if acknowledge:
## acknoldge invalidation in mcm, provided we can have the api
print "acknowledgment to mcm"
mcm.get('/restapi/invalidations/acknowledge/%s' % (invalid['_id']))
# prepare the text for batches
batches = []
batches.extend(
mcm.getA('batches', query='contains=%s' % batch_lookup))
batches = filter(
lambda b: b['status'] in ['announced', 'done', 'reset'],
batches)
if len(batches):
bid = batches[-1]['prepid']
print "batch nofication to", bid
text_to_batch[bid] += text + "\n\n"
# prepare the text for requests
text_to_request[pid] += text + "\n\n"
for bid, text in text_to_batch.items():
if not text: continue
text += '\n This is an automated message'
mcm.put('/restapi/batches/notify', {"notes": text, "prepid": bid})
pass
for pid, text in text_to_request.items():
if not text: continue
text += '\n This is an automated message'
mcm.put('/restapi/requests/notify', {
"message": text,
"prepids": [pid]
})
def rejector(url, specific, options=None):
up = componentInfo()
if specific and specific.startswith('/'):
return
if options.filelist:
wfs = []
for line in filter(None, open(options.filelist).read().split('\n')):
print line
wfs.extend(
session.query(Workflow).filter(
Workflow.name.contains(line)).all())
elif specific:
wfs = session.query(Workflow).filter(
Workflow.name.contains(specific)).all()
else:
return
print len(wfs), "to reject"
if len(wfs) > 1:
print "\n".join([wfo.name for wfo in wfs])
answer = raw_input('Reject these')
if not answer.lower() in ['y', 'yes']:
return
for wfo in wfs:
#wfo = session.query(Workflow).filter(Workflow.name == specific).first()
if not wfo:
print "cannot reject", spec
return
results = []
wfi = workflowInfo(url, wfo.name)
datasets = set(wfi.request['OutputDatasets'])
reqMgrClient.invalidateWorkflow(
url, wfo.name, current_status=wfi.request['RequestStatus'])
wfi.sendLog('rejector',
'invalidating the workflow by unified operator')
## need to find the whole familly and reject the whole gang
familly = getWorkflowById(url, wfi.request['PrepID'], details=True)
for fwl in familly:
if fwl['RequestDate'] < wfi.request['RequestDate']: continue
if fwl['RequestType'] != 'Resubmission': continue
if 'OriginalRequestName' in fwl and fwl[
'OriginalRequestName'] != wfi.request['RequestName']:
continue
print "rejecting", fwl['RequestName']
reqMgrClient.invalidateWorkflow(
url, fwl['RequestName'], current_status=fwl['RequestStatus'])
datasets.update(fwl['OutputDatasets'])
for dataset in datasets:
if options.keep:
print "keeping", dataset, "in its current status"
else:
results.append(setDatasetStatus(dataset, 'INVALID'))
#if wfi.request['RequestStatus'] in ['rejected','rejected-archived','aborted','aborted-archived']:
# print 'already',wfi.request['RequestStatus']
# if not options.clone:
# wfo.status = 'forget'
# session.commit()
# continue
if all(map(lambda result: result in ['None', None, True], results)):
wfo.status = 'forget'
session.commit()
print wfo.name, "and", datasets, "are rejected"
if options and options.clone:
schema = wfi.getSchema()
schema['Requestor'] = os.getenv('USER')
schema['Group'] = 'DATAOPS'
schema['OriginalRequestName'] = wfo.name
if 'ProcessingVersion' in schema:
schema['ProcessingVersion'] = int(
schema['ProcessingVersion']
) + 1 ## dubious str->int conversion
else:
schema['ProcessingVersion'] = 2
for k in schema.keys():
if k.startswith('Team'):
schema.pop(k)
if k.startswith('checkbox'):
schema.pop(k)
## a few tampering of the original request
if options.Memory:
schema['Memory'] = options.Memory
if options.deterministic:
if schema['RequestType'] == 'TaskChain':
schema['Task1']['DeterministicPileup'] = True
if options.EventsPerJob:
if schema['RequestType'] == 'TaskChain':
schema['Task1']['EventsPerJob'] = options.EventsPerJob
else:
schema['EventsPerJob'] = options.EventsPerJob
if options.EventAwareLumiBased:
schema['SplittingAlgo'] = 'EventAwareLumiBased'
if options.TimePerEvent:
schema['TimePerEvent'] = options.TimePerEvent
if options.ProcessingString:
schema['ProcessingString'] = options.ProcessingString
if options.AcquisitionEra:
schema['AcquisitionEra'] = options.AcquisitionEra
if options.runs:
schema['RunWhitelist'] = map(int, options.runs.split(','))
if options.PrepID:
schema['PrepID'] = options.PrepID
if schema['RequestType'] == 'TaskChain' and options.no_output:
ntask = schema['TaskChain']
for it in range(1, ntask - 1):
schema['Task%d' % it]['KeepOutput'] = False
schema['TaskChain'] = ntask - 1
schema.pop('Task%d' % ntask)
## update to the current priority
schema['RequestPriority'] = wfi.request['RequestPriority']
## drop shit on the way to reqmgr2
for p in [
'RequestStatus',
'RequestTransition',
'RequestorDN',
'RequestWorkflow',
'OutputDatasets',
'ReqMgr2Only',
#'Group',
'RequestDate',
#'ConfigCacheUrl',
'RequestName',
'timeStamp',
'SoftwareVersions',
'CouchURL'
]:
if p in schema:
schema.pop(p)
#pass
print "submitting"
print json.dumps(schema, indent=2)
newWorkflow = reqMgrClient.submitWorkflow(url, schema)
if not newWorkflow:
print "error in cloning", wfo.name
print json.dumps(schema, indent=2)
return
print newWorkflow
#m = re.search("details\/(.*)\'",response)
#if not m:
# print "error in cloning",wfo.name
# print response
# print json.dumps( schema, indent=2 )
# return
#newWorkflow = m.group(1)
data = reqMgrClient.setWorkflowApproved(url, newWorkflow)
print data
wfo.status = 'trouble'
session.commit()
wfi.sendLog(
'rejector',
'Cloned into %s by unified operator' % (newWorkflow))
else:
print "error in rejecting", wfo.name, results
def rejector(url, specific, options=None):
up = componentInfo()
if specific and specific.startswith('/'):
return
if options.filelist:
wfs = []
for line in filter(None, open(options.filelist).read().split('\n')):
print line
wfs.extend( session.query(Workflow).filter(Workflow.name.contains(line)).all())
elif specific:
wfs = session.query(Workflow).filter(Workflow.name.contains(specific)).all()
else:
return
print len(wfs),"to reject"
if len(wfs)>1:
print "\n".join( [wfo.name for wfo in wfs] )
answer = raw_input('Reject these')
if not answer.lower() in ['y','yes']:
return
for wfo in wfs:
#wfo = session.query(Workflow).filter(Workflow.name == specific).first()
if not wfo:
print "cannot reject",spec
return
results=[]
wfi = workflowInfo(url, wfo.name)
datasets = set(wfi.request['OutputDatasets'])
reqMgrClient.invalidateWorkflow(url, wfo.name, current_status=wfi.request['RequestStatus'])
wfi.sendLog('rejector','invalidating the workflow by unified operator')
## need to find the whole familly and reject the whole gang
familly = getWorkflowById( url, wfi.request['PrepID'] , details=True)
for fwl in familly:
if fwl['RequestDate'] < wfi.request['RequestDate']: continue
if fwl['RequestType']!='Resubmission': continue
if 'OriginalRequestName' in fwl and fwl['OriginalRequestName'] != wfi.request['RequestName']: continue
print "rejecting",fwl['RequestName']
reqMgrClient.invalidateWorkflow(url, fwl['RequestName'], current_status=fwl['RequestStatus'])
datasets.update( fwl['OutputDatasets'] )
for dataset in datasets:
if options.keep:
print "keeping",dataset,"in its current status"
else:
results.append( setDatasetStatus(dataset, 'INVALID') )
#if wfi.request['RequestStatus'] in ['rejected','rejected-archived','aborted','aborted-archived']:
# print 'already',wfi.request['RequestStatus']
# if not options.clone:
# wfo.status = 'forget'
# session.commit()
# continue
if all(map(lambda result : result in ['None',None,True],results)):
wfo.status = 'forget'
session.commit()
print wfo.name,"and",datasets,"are rejected"
if options and options.clone:
schema = wfi.getSchema()
schema['Requestor'] = os.getenv('USER')
schema['Group'] = 'DATAOPS'
schema['OriginalRequestName'] = wfo.name
if 'ProcessingVersion' in schema:
schema['ProcessingVersion'] = int(schema['ProcessingVersion'])+1 ## dubious str->int conversion
else:
schema['ProcessingVersion']=2
for k in schema.keys():
if k.startswith('Team'):
schema.pop(k)
if k.startswith('checkbox'):
schema.pop(k)
## a few tampering of the original request
if options.Memory:
schema['Memory'] = options.Memory
if options.deterministic:
if schema['RequestType'] == 'TaskChain':
schema['Task1']['DeterministicPileup'] = True
if options.EventsPerJob:
if schema['RequestType'] == 'TaskChain':
schema['Task1']['EventsPerJob'] = options.EventsPerJob
else:
schema['EventsPerJob'] = options.EventsPerJob
if options.EventAwareLumiBased:
schema['SplittingAlgo'] = 'EventAwareLumiBased'
if options.TimePerEvent:
schema['TimePerEvent'] = options.TimePerEvent
if options.ProcessingString:
schema['ProcessingString'] = options.ProcessingString
if options.AcquisitionEra:
schema['AcquisitionEra'] = options.AcquisitionEra
if options.runs:
schema['RunWhitelist'] = map(int,options.runs.split(','))
if options.PrepID:
schema['PrepID'] =options.PrepID
if schema['RequestType'] == 'TaskChain' and options.no_output:
ntask = schema['TaskChain']
for it in range(1,ntask-1):
schema['Task%d'%it]['KeepOutput'] = False
schema['TaskChain'] = ntask-1
schema.pop('Task%d'%ntask)
## update to the current priority
schema['RequestPriority'] = wfi.request['RequestPriority']
## drop shit on the way to reqmgr2
for p in ['RequestStatus',
'RequestTransition',
'RequestorDN',
'RequestWorkflow',
'OutputDatasets',
'ReqMgr2Only',
#'Group',
'RequestDate',
#'ConfigCacheUrl',
'RequestName',
'timeStamp',
'SoftwareVersions',
'CouchURL'
]:
if p in schema:
schema.pop( p )
#pass
print "submitting"
print json.dumps( schema, indent=2 )
newWorkflow = reqMgrClient.submitWorkflow(url, schema)
if not newWorkflow:
print "error in cloning",wfo.name
print json.dumps( schema, indent=2 )
return
print newWorkflow
#m = re.search("details\/(.*)\'",response)
#if not m:
# print "error in cloning",wfo.name
# print response
# print json.dumps( schema, indent=2 )
# return
#newWorkflow = m.group(1)
data = reqMgrClient.setWorkflowApproved(url, newWorkflow)
print data
wfo.status = 'trouble'
session.commit()
wfi.sendLog('rejector','Cloned into %s by unified operator'%( newWorkflow ))
else:
print "error in rejecting",wfo.name,results
def rejector(url, specific, options=None):
up = componentInfo(soft=['wtc'])
if not up.check(): return
if specific and specific.startswith('/'):
## this is for a dataset
print setDatasetStatus(specific, 'INVALID')
return
if options.filelist:
wfs = []
for line in filter(None, open(options.filelist).read().split('\n')):
print line
wfs.extend(
session.query(Workflow).filter(
Workflow.name.contains(line)).all())
elif specific:
wfs = session.query(Workflow).filter(
Workflow.name.contains(specific)).all()
if not wfs:
batches = json.loads(open('batches.json').read())
for bname in batches:
if specific == bname:
for wf in batches[bname]:
wfs.append(
session.query(Workflow).filter(
Workflow.name == wf).first())
else:
wfs = session.query(Workflow).filter(
Workflow.status == 'assistance-clone').all()
#wfs.extend( session.query(Workflow).filter(Workflow.status == 'assistance-reject').all())
## be careful then on clone case by case
options.clone = True
print "not supposed to function yet"
return
print len(wfs), "to reject"
if len(wfs) > 1:
print "\n".join([wfo.name for wfo in wfs])
answer = raw_input('Reject these')
if not answer.lower() in ['y', 'yes']:
return
for wfo in wfs:
#wfo = session.query(Workflow).filter(Workflow.name == specific).first()
if not wfo:
print "cannot reject", spec
return
results = []
wfi = workflowInfo(url, wfo.name)
datasets = set(wfi.request['OutputDatasets'])
reqMgrClient.invalidateWorkflow(
url, wfo.name, current_status=wfi.request['RequestStatus'])
comment = ""
if options.comments: comment = ", reason: " + options.comments
wfi.sendLog(
'rejector',
'invalidating the workflow by unified operator%s' % comment)
## need to find the whole familly and reject the whole gang
familly = getWorkflowById(url, wfi.request['PrepID'], details=True)
for fwl in familly:
if fwl['RequestDate'] < wfi.request['RequestDate']: continue
if fwl['RequestType'] != 'Resubmission': continue
## does not work on second order acd
#if 'OriginalRequestName' in fwl and fwl['OriginalRequestName'] != wfi.request['RequestName']: continue
print "rejecting", fwl['RequestName']
reqMgrClient.invalidateWorkflow(
url,
fwl['RequestName'],
current_status=fwl['RequestStatus'],
cascade=False)
datasets.update(fwl['OutputDatasets'])
for dataset in datasets:
if options.keep:
print "keeping", dataset, "in its current status"
else:
results.append(setDatasetStatus(dataset, 'INVALID'))
pass
if all(map(lambda result: result in ['None', None, True], results)):
print wfo.name, "and", datasets, "are rejected"
if options and options.clone:
wfo.status = 'trouble'
session.commit()
schema = wfi.getSchema()
schema['Requestor'] = os.getenv('USER')
schema['Group'] = 'DATAOPS'
schema['OriginalRequestName'] = wfo.name
if 'ProcessingVersion' in schema:
schema['ProcessingVersion'] = int(
schema['ProcessingVersion']
) + 1 ## dubious str->int conversion
else:
schema['ProcessingVersion'] = 2
for k in schema.keys():
if k.startswith('Team'):
schema.pop(k)
if k.startswith('checkbox'):
schema.pop(k)
## a few tampering of the original request
if options.Memory:
if schema['RequestType'] == 'TaskChain':
it = 1
while True:
t = 'Task%d' % it
it += 1
if t in schema:
schema[t]['Memory'] = options.Memory
else:
break
else:
schema['Memory'] = options.Memory
if options.Multicore:
## to do : set it properly in taskchains
if schema['RequestType'] == 'TaskChain':
tasks, set_to = options.Multicore.split(
':') if ':' in options.Multicore else (
"", options.Multicore)
set_to = int(set_to)
tasks = tasks.split(',') if tasks else ['Task1']
it = 1
while True:
tt = 'Task%d' % it
it += 1
if tt in schema:
tname = schema[tt]['TaskName']
if tname in tasks or tt in tasks:
mem = schema[tt]['Memory']
mcore = schema[tt].get('Multicore', 1)
factor = (set_to / float(mcore))
fraction_constant = 0.4
mem_per_core_c = int(
(1 - fraction_constant) * mem /
float(mcore))
print "mem per core", mem_per_core_c
print "base mem", mem
## adjusting the parameter in the clone
schema[tt]['Memory'] += (
set_to - mcore) * mem_per_core_c
schema[tt]['Multicore'] = set_to
schema[tt]['TimePerEvent'] /= factor
else:
break
else:
schema['Multicore'] = options.Multicore
if options.deterministic:
if schema['RequestType'] == 'TaskChain':
schema['Task1']['DeterministicPileup'] = True
if options.EventsPerJob:
if schema['RequestType'] == 'TaskChain':
schema['Task1']['EventsPerJob'] = options.EventsPerJob
else:
schema['EventsPerJob'] = options.EventsPerJob
if options.EventAwareLumiBased:
schema['SplittingAlgo'] = 'EventAwareLumiBased'
if options.TimePerEvent:
schema['TimePerEvent'] = options.TimePerEvent
if options.ProcessingString:
schema['ProcessingString'] = options.ProcessingString
if options.AcquisitionEra:
schema['AcquisitionEra'] = options.AcquisitionEra
if options.runs:
schema['RunWhitelist'] = map(int, options.runs.split(','))
if options.PrepID:
schema['PrepID'] = options.PrepID
if schema['RequestType'] == 'TaskChain' and options.no_output:
ntask = schema['TaskChain']
for it in range(1, ntask - 1):
schema['Task%d' % it]['KeepOutput'] = False
schema['TaskChain'] = ntask - 1
schema.pop('Task%d' % ntask)
if options.priority:
schema['RequestPriority'] = options.priority
## update to the current priority
schema['RequestPriority'] = wfi.request['RequestPriority']
## drop shit on the way to reqmgr2
schema = reqMgrClient.purgeClonedSchema(schema)
print "submitting"
if (options.to_stepchain
and (schema['RequestType'] == 'TaskChain')):
## transform the schema into StepChain schema
print "Transforming a TaskChain into a StepChain"
schema['RequestType'] = 'StepChain'
schema['StepChain'] = schema.pop('TaskChain')
schema['SizePerEvent'] = 0
schema['TimePerEvent'] = 0
step = 1
s_n = {}
while True:
if 'Task%d' % step in schema:
schema['Step%d' % step] = schema.pop('Task%d' %
step)
schema['Step%d' % step]['StepName'] = schema[
'Step%d' % step].pop('TaskName')
s_n[schema['Step%d' %
step]['StepName']] = 'Step%d' % step
if 'InputTask' in schema['Step%d' % step]:
schema['Step%d' % step]['InputStep'] = schema[
'Step%d' % step].pop('InputTask')
eff = 1.
up_s = 'Step%d' % step
while True:
## climb up a step. supposedely already all converted
up_s = s_n.get(
schema[up_s].get('InputStep', None), None)
if up_s:
## multiply with the efficiency
eff *= schema[up_s].get(
'FilterEfficiency', 1.)
else:
## or stop there
break
if not 'KeepOutput' in schema['Step%d' % step]:
## this is a weird translation capability. Absence of keepoutput in step means : keep the output. while in TaskChain absence means : drop
schema['Step%d' % step]['KeepOutput'] = False
schema['TimePerEvent'] += eff * schema[
'Step%d' % step].pop('TimePerEvent')
schema['SizePerEvent'] += eff * schema[
'Step%d' % step].pop('SizePerEvent')
step += 1
else:
break
print json.dumps(schema, indent=2)
newWorkflow = reqMgrClient.submitWorkflow(url, schema)
if not newWorkflow:
print "error in cloning", wfo.name
print json.dumps(schema, indent=2)
return
print newWorkflow
data = reqMgrClient.setWorkflowApproved(url, newWorkflow)
print data
wfi.sendLog(
'rejector', 'Cloned into %s by unified operator %s' %
(newWorkflow, comment))
wfi.notifyRequestor('Cloned into %s by unified operator %s' %
(newWorkflow, comment),
do_batch=False)
else:
wfo.status = 'trouble' if options.set_trouble else 'forget'
wfi.notifyRequestor('Rejected by unified operator %s' %
(comment),
do_batch=False)
session.commit()
else:
print "error in rejecting", wfo.name, results
def main():
usage = (
"\n python %prog [options] [WORKFLOW_NAME]\n"
"WORKFLOW_NAME: if the list file is provided this should be empty\n"
)
parser = OptionParser(usage=usage)
parser.add_option("-f", "--file", help="Text file of workflows to Reject and Clone", dest="file")
parser.add_option(
"-c",
"--clone",
help="Are the workflows going to be cloned? The default value is False",
action="store_true",
dest="clone",
default=False,
)
parser.add_option(
"-i",
"--invalidate",
help="Invalidate datasets? The default value is False",
action="store_true",
dest="invalidate",
default=False,
)
parser.add_option(
"-u",
"--user",
dest="user",
help="The user for creating the clone, if empty it will use the OS user running the script",
)
parser.add_option(
"-g",
"--group",
dest="group",
default="DATAOPS",
help="The group for creating the clone, if empty it will, use 'DATAOPS' by default",
)
parser.add_option(
"-m",
"--memory",
dest="memory",
help="Set max memory for the clone. At assignment, this will be used to calculate maxRSS = memory*1024",
)
(options, args) = parser.parse_args()
# Check the arguments, get info from them
if options.file:
try:
workflows = [l.strip() for l in open(options.file) if l.strip()]
except:
parser.error("Provide a valid file of workflows")
sys.exit(1)
elif len(args) > 0:
# name of workflow
workflows = [args[0]]
else:
parser.error("Provide the workflow of a file of workflows")
sys.exit(1)
if not options.user:
# get os username by default
uinfo = pwd.getpwuid(os.getuid())
user = uinfo.pw_name
else:
user = options.user
for workflow in workflows:
try:
workflowInfo = reqMgrClient.Workflow(workflow)
except:
print("The workflow name: " + workflow + " is not valid.")
continue
# invalidates workflow
print("Invalidating the workflow: " + workflow)
reqMgrClient.invalidateWorkflow(url, workflow, workflowInfo.status)
# invalidates datasets
if options.invalidate:
print("Invalidating datasets")
datasets = reqMgrClient.outputdatasetsWorkflow(url, workflow)
for dataset in datasets:
print(dataset)
dbs3.setDatasetStatus(dataset, "INVALID", files=True)
# clones workflow
if options.clone:
print("Cloning workflow: " + workflow)
if options.memory:
mem = float(options.memory)
else:
mem = workflowInfo.info["Memory"]
cloned = resubmit.cloneWorkflow(workflow, user, options.group, memory=mem)
sys.exit(0)
