Python utils.workflowInfo函数代码示例

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 invalidator(url, invalid_status='INVALID'):
    mcm = McMClient(dev=False)
    invalids = mcm.getA('invalidations',query='status=new')
    print len(invalids),"Object to be invalidated"
    for invalid in invalids:
        acknowledge= False
        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)
                wfo.status = 'forget'
                session.commit()
            wfi = workflowInfo(url, wfn)
            success = "not rejected"
            if wfi.request['RequestStatus'] in ['assignment-approved','new','completed']:
                success = reqMgrClient.rejectWorkflow(url, wfn)
                pass
            else:
                success = reqMgrClient.abortWorkflow(url, wfn)
                pass
            print success
            acknowledge= True
        elif invalid['type'] == 'dataset':
            dataset = invalid['object']
            if 'None-' in dataset: continue
            if 'FAKE-' in dataset: continue
            print "setting",dataset,"to",invalid_status
            success = "not invalidated"
            success = setDatasetStatusDBS3.setStatusDBS3('https://cmsweb.cern.ch/dbs/prod/global/DBSWriter', dataset, invalid_status, None)
            print success
            ## make a delete request from everywhere we can find ?
            acknowledge= True
        else:
            print "\t\t",invalid['type']," type not recognized"

        if acknowledge:
            ## acknoldge invalidation in mcm, provided we can have the api
            print "No acknowledgment api yet available"

def releasor():
    if duplicateLock() : return

    SI = siteInfo()
    CI = campaignInfo()
    LI = lockInfo()

    tiers_no_custodial = ['MINIADOSIM']
    wfs = []
    for fetch in ['done','forget']:
        wfs.extend( session.query(Workflow).filter(Workflow.status==fetch).all() )
    
    
    for wfo in wfs:
        wfi = workflowInfo(url, wfo.name )
        announced_log = filter(lambda change : change["Status"] in ["closed-out","normal-archived","announced"],wfi.request['RequestTransition'])
        if not announced_log: 
            print "Cannot figure out when",wfo.name,"was finished"
            continue
        now = time.mktime(time.gmtime()) / (60*60*24.)
        then = announced_log[-1]['UpdateTime'] / (60.*60.*24.)
        if (now-then) <2:
            print "workflow",wfo.name, "finished",now-then,"days ago. Too fresh to clean"
            continue
        else:
            print "workflow",wfo.name,"has finished",now-then,"days ago."

        (_,primaries,_,secondaries) = wfi.getIO()
        outputs = wfi.request['OutputDatasets']
        datasets_to_check = list(primaries)+list(secondaries)+outputs
        for dataset in datasets_to_check:
            (_,_,_,tier) = dataset.split('/')
            ## check custodial if required
            if tier not in tiers_no_custodial:
            ## check not used anymore by anything
            ## unlock output and input everywhere if so
            pass

def getPriority(url, workflow): 
	return workflowInfo( url,workflow).getPriority()

def singleClone(url, wfname, actions, comment, do=False):
    
    wfi = workflowInfo(url, wfname)
    payload = wfi.getSchema()
    initial = wfi.request

    payload['Requestor']           = os.getenv('USER')
    payload['Group']               = 'DATAOPS'
    payload['OriginalRequestName'] = initial['RequestName']
    payload['RequestPriority'] = initial['RequestPriority']

    if 'ProcessingVersion' in initial:
        payload['ProcessingVersion'] = int(initial['ProcessingVersion']) +1
    else:
        payload['ProcessingVersion'] = 2


## drop parameters 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','checkbox',
                  'DN', 'AutoApproveSubscriptionSites', 'NonCustodialSites', 'CustodialSites', 'OriginalRequestName', 'Teams', 'OutputModulesLFNBases', 
                  'SiteBlacklist', 'AllowOpportunistic', '_id']
    for p in paramBlacklist:
        if p in payload:
            payload.pop( p )
            pass

    if actions:
        for action in actions:
            if action.startswith('mem') and actions[action] != "" and actions[action] != 'Same':
                if 'TaskChain' in payload:
                    print "Setting memory for clone of task chain"
                    it=1
                    while True:
                        t = 'Task%d'%it
                        it+=1
                        if t in payload:
                            payload[t]['Memory'] = actions[action]
                            print "Memory set for Task%d"%it
                        else:
                            break
                else:
                    print "Setting memory for non-taskchain workflow"
                    payload['Memory'] = actions[action]
                print "Memory set to " + actions[action]
                #This line is doesn't work for some reason
#                wfi.sendLog('actor','Memory of clone set to %d'%actions[action])

    print "Clone payload"
#    print json.dumps( payload , indent=2)
    print actions

    #Create clone
    clone = reqMgrClient.submitWorkflow(url, payload)
    if not clone:
        print "Error in making clone for",initial["RequestName"]
        clone = reqMgrClient.submitWorkflow(url, payload)
        if not clone:
            print "Error twice in making clone for",initial["RequestName"]
            sendLog('actor','Failed to make a clone twice for %s!'%initial["RequestName"],level='critical')
            wfi.sendLog('actor','Failed to make a clone twice for %s!'%initial["RequestName"])
            return None

    if actions:
        for action in actions:
            if action.startswith('split'):
                cloneinfo = workflowInfo(url, clone)
                splittings = cloneinfo.getSplittings()
                if actions[action] != 'Same' and actions[action] != 'max' and actions[action] != '':
                    factor = int(actions[action][0:-1]) if 'x' in actions[action] else 2
                    for split in splittings:
                        for act in ['avg_events_per_job','events_per_job','lumis_per_job']:
                            if act in split:
                                wfi.sendLog('actor','Changing %s (%d) by a factor %d'%( act, split[act], factor))
                                print "Changing %s (%d) by a factor %d"%( act, split[act], factor),
                                split[act] /= factor
                                print "to",split[act]
                                break
                        split['requestName'] = clone
                        print "changing the splitting of",clone
                        print json.dumps( split, indent=2 )
                        print reqMgrClient.setWorkflowSplitting(url, clone, split )
                elif 'max' in actions[action]:
                    for split in splittings:
                        for act in ['avg_events_per_job','events_per_job','lumis_per_job']:
                            if act in split:
                                wfi.sendLog('actor','Max splitting set for %s (%d'%( act, split[act]))
                                print "Changing %s (%d) "%( act, split[act]),
                                split[act] = 1
                                print "to max splitting ",split[act]
                                break
                        split['requestName'] = clone
                        print "changing the splitting of",clone
                        print json.dumps( split, indent=2 )
#.........这里部分代码省略.........

wfs_no_location_in_GQ = defaultdict(list)
si = siteInfo()  
#bad_blocks = defaultdict( set )
unprocessable = set()

not_runable_acdc=set()
agents_down = defaultdict(set)
failed_workflow = set()
files_locations = {}
stuck_all_done = set()
heavy_duty = {}

for wf in wfs:
    if spec and not spec in wf['RequestName']: continue

    wfi = workflowInfo(url, wf['RequestName'], request=wf)
    sitewhitelist = wfi.request['SiteWhitelist']
    wqs = wfi.getWorkQueue()
    
    stats = wfi.getWMStats()
    if not 'AgentJobInfo' in stats: stats['AgentJobInfo'] = {}

    ## skip wf that unified does not know about, leaves acdc
    wfo = session.query(Workflow).filter(Workflow.name == wf['RequestName']).first()
    if not (wfo or wf['RequestType']=='Resubmission'): 
        print "not knonw or not acdc : %s"%(wf['RequestName'])
        continue

    ## test the heavyness
    if 'TotalInputLumis' in wf and 'TotalEstimatedJobs' in wf and wf['TotalEstimatedJobs']:
        heavy = (wf['TotalInputLumis'] / float(wf['TotalEstimatedJobs']))

def outcleanor(url, options):

    if options.approve:
        for user in ['*Vlimant']:#,'*Cremonesi']:
            deletes = listDelete( url , user = user)
            for (site,who,tid) in deletes:
                if 'MSS' in site: continue### ever
                print site,who,tid
                print "approving deletion"
                print approveSubscription(url, tid, nodes = [site], comments = 'Production cleaning by data ops')
        return

    

    sites_and_datasets = defaultdict(list)
    our_copies = defaultdict(list)
    wf_cleaned = {}
    
    wfs = []
    for fetch in options.fetch.split(','):
        wfs.extend(session.query(Workflow).filter(Workflow.status==fetch).all())

    random.shuffle( wfs )
    last_answer = None
    for wfo in wfs :
        if options.number and len(wf_cleaned)>= options.number:
            print "Reached",options.number,"cleaned"
            break
        print '-'*100
        wfi = workflowInfo(url, wfo.name)
        goes = {} # boolean per output
        for dataset in wfi.request['OutputDatasets']:
            goes[dataset] = False
            keep_one_out = True
            status = getDatasetStatus( dataset )
            print "\n\tLooking at",dataset,status,"\n"
            vetoes = None
            if status == 'INVALID':
                vetoes = ['Export','Buffer'] ## can take themselves out
                keep_one_out = False # just wipe clean

            elif status == None:
                print dataset,"actually does not exist. skip"
                goes[dataset] = True
                continue

            elif status in ['PRODUCTION','VALID'] and wfo.status in ['forget','trouble']:
                print dataset,"should probably be invalidated. (",wfo.status,") skip"
                keep_one_out = False # just wipe clean
                continue ## you are not sure. just skip it for the time being

            elif status == 'PRODUCTION' and wfo.status in ['clean']:
                print dataset,"should probably be set valid .skip"
                continue ## you are not sure. just skip it for the time being

            if status == 'VALID' and dataset.startswith('/MinBias'):
                print "This is a /MinBias. skip"
                continue

            if '/DQM' in dataset:
                keep_one_out = False

            total_size = getDatasetSize( dataset )
            
            our_presence = getDatasetPresence(url, dataset, complete=None, group="DataOps", vetoes=vetoes)
            also_our_presence = getDatasetPresence(url, dataset, complete=None, group="", vetoes=vetoes)
            
            ## merge in one unique dict
            for site in also_our_presence:
                if site in our_presence:
                    there,frac = our_presence[site]
                    other,ofrac = also_our_presence[site]
                    our_presence[site] = (max(there,other),max(frac,ofrac))
                else:
                    our_presence[site] = also_our_presence[site]
                
            if our_presence: print our_presence

            ## analysis ops copies need to be taken into account
            anaops_presence = getDatasetPresence(url, dataset, complete=None, group="AnalysisOps")
            own_by_anaops = anaops_presence.keys()
            
            ## all our copies
            to_be_cleaned = our_presence.keys()
            if not len(to_be_cleaned):
                print "nowhere to be found of ours,",len(own_by_anaops),"in analysi ops pool"
                goes[dataset] = True
                continue

            print "Where we own bits of dataset"
            print to_be_cleaned
     

            if len(own_by_anaops):
                ## remove site with the anaops copies
                to_be_cleaned = list(set(to_be_cleaned) - set(own_by_anaops))
                keep_one_out = False ## in that case, just remove our copies
                print "Own by anaops (therefore not keep a copy of ours)"
                print own_by_anaops
            else:
#.........这里部分代码省略.........

def recoveror(url,specific,options=None):
    if userLock('recoveror'): return

    up = componentInfo()
    CI = campaignInfo()

    UC = unifiedConfiguration()

    def make_int_keys( d ):
        for code in d:
            d[int(code)] = d.pop(code)

    error_codes_to_recover = UC.get('error_codes_to_recover')
    error_codes_to_block = UC.get('error_codes_to_block')
    error_codes_to_notify = UC.get('error_codes_to_notify')
    make_int_keys( error_codes_to_recover )
    make_int_keys( error_codes_to_block )
    make_int_keys( error_codes_to_notify )

    wfs = session.query(Workflow).filter(Workflow.status == 'assistance-recovery').all()
    if specific:
        wfs.extend( session.query(Workflow).filter(Workflow.status == 'assistance-manual').all() )

    for wfo in wfs:
        if specific and not specific in wfo.name:continue

        if not specific and 'manual' in wfo.status: continue
        
        wfi = workflowInfo(url, wfo.name, deprecated=True) ## need deprecated info for mergedlfnbase

        ## 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

        all_errors = None
        try:
            wfi.getSummary()
            all_errors = wfi.summary['errors']
        except:
            pass

        print '-'*100        
        print "Looking at",wfo.name,"for recovery options"
        
        if not len(all_errors): 
            print "\tno error for",wfo.name

        task_to_recover = defaultdict(list)
        message_to_ops = ""
        message_to_user = ""

        recover=True
        if 'LheInputFilese' in wfi.request and wfi.request['LheInputFiles']:
            ## we do not try to recover pLHE
            recover = False

        if 'Campaign' in wfi.request:
            c = wfi.request['Campaign']
            if c in CI.campaigns and 'recover' in CI.campaigns[c]:
                recover=CI.campaigns[c]['recover']

        for task,errors in all_errors.items():
            print "\tTask",task
            ## collect all error codes and #jobs regardless of step at which it occured
            all_codes = []
            for name, codes in errors.items():
                if type(codes)==int: continue
                all_codes.extend( [(int(code),info['jobs'],name,list(set([e['type'] for e in info['errors']])),list(set([e['details'] for e in info['errors']])) ) for code,info in codes.items()] )

            all_codes.sort(key=lambda i:i[1], reverse=True)
            sum_failed = sum([l[1] for l in all_codes])

            for errorCode,njobs,name,types,details in all_codes:
                rate = 100*njobs/float(sum_failed)
                #print ("\t\t %10d (%6s%%) failures with error code %10d (%"+str(max_legend)+"s) at stage %s")%(njobs, "%4.2f"%rate, errorCode, legend, name)
                print ("\t\t %10d (%6s%%) failures with error code %10d (%30s) at stage %s")%(njobs, "%4.2f"%rate, errorCode, ','.join(types), name)
                    
                added_in_recover=False

                #if options.go:
                # force the recovery of any task with error ?

                if errorCode in error_codes_to_recover:
                    ## the error code is registered
                    for case in error_codes_to_recover[errorCode]:
                        match = case['details']
                        matched= (match==None)
                        if not matched:
                            matched=False
                            for detail in details:
                                if match in detail:
                                    print "[recover] Could find keyword",match,"in"
                                    print 50*"#"
                                    print detail
                                    print 50*"#"
                                    matched = True
                                    break
                        if matched and rate > case['rate']:
                            print "\t\t => we should be able to recover that", case['legend']
                            task_to_recover[task].append( (code,case) )
                            added_in_recover=True
                            message_to_user = ""
#.........这里部分代码省略.........

def main():

    #Create option parser
    usage = "usage: %prog (-w workflow|-f filelist) (-t TASK|--all) [--tesbed]"
    parser = OptionParser(usage=usage)
    parser.add_option("-f","--file", dest="file", default=None,
                        help="Text file with a list of workflows")
    parser.add_option("-w","--workflow", default=None,
                      help="Coma separated list of wf to handle")
    parser.add_option("-t","--task", default=None,
                      help="Coma separated task to be recovered")
    parser.add_option("-p","--path", default=None,
                      help="Coma separated list of paths to recover")
    parser.add_option("-a","--all",
                      help="Make acdc for all tasks to be recovered",default=False, action='store_true')
    parser.add_option("-m","--memory", dest="memory", default=None, type=int,
                        help="Memory to override the original request memory")
    parser.add_option("-c","--mcore", dest="mcore", default=None,
                      help="Multicore to override the original request multicore")
    parser.add_option("--testbed", default=False, action="store_true")

    (options, args) = parser.parse_args()

    global url
    url = testbed_url if options.testbed else prod_url

    if options.all : options.task = 'all'

    if not options.task:
        parser.error("Provide the -t Task Name or --all")
        sys.exit(1)

    if not ((options.workflow) or (options.path) or (options.file)):
        parser.error("Provide the -w Workflow Name or the -p path or the -f workflow filelist")
        sys.exit(1)
    
    wfs = None
    wf_and_task = defaultdict(set)
    if options.file:
        wfs = [l.strip() for l in open(options.file) if l.strip()]
    elif options.workflow:
        wfs = options.workflow.split(',')
    elif options.path:
        ## self contained
        paths = options.path.split(',')
        for p in paths:
            _,wf,t = p.split('/',2)
            wf_and_task[wf].add('/%s/%s'%(wf,t))
    else:
        parser.error("Either provide a -f filelist or a -w workflow or -p path")
        sys.exit(1)

    if not wf_and_task:
        if options.task == 'all':
            for wfname in wfs: 
                wf_and_task[wfname] = None
        else:
            for wfname in wfs: 
                wf_and_task[wfname].update( [('/%s/%s'%(wfname,task)).replace('//','/') for task in options.task.split(',')] )

    if not wf_and_task:
        parser.error("Provide the -w Workflow Name and the -t Task Name or --all")
        sys.exit(1)        

    
    for wfname,tasks in wf_and_task.items():
        wfi = workflowInfo(url, wfname)
        if tasks == None:
            where,how_much,how_much_where = wfi.getRecoveryInfo()
            tasks = sorted(how_much.keys())
        else:
            tasks = sorted(tasks)

        created = {}
        print "Workflow:",wfname
        print "Tasks:",tasks
        for task in tasks:
            r = makeACDC(url=url, wfi=wfi, task=task,
                         memory = options.memory,
                         mcore = options.mcore) 
            if not r: 
                print "Error in creating ACDC for",task,"on",wfname
                break
            created[task] = r
        if len(created)!=len(tasks):
            print "Error in creating all required ACDCs"
            sys.exit(1)
        print "Created:"
        for task in created:
            print created[task],"for",task

def assignor(url ,specific = None, talk=True, options=None):
    if userLock(): return
    if duplicateLock(): return
    if not componentInfo().check(): return

    UC = unifiedConfiguration()
    CI = campaignInfo()
    #SI = siteInfo()
    SI = global_SI()
    #NLI = newLockInfo()
    #if not NLI.free() and not options.go: return
    LI = lockInfo()
    if not LI.free() and not options.go: return

    n_assigned = 0
    n_stalled = 0

    wfos=[]
    fetch_from = []
    if specific or options.early:
        fetch_from.extend(['considered','staging'])
    if specific:
        fetch_from.extend(['considered-tried'])
    
    fetch_from.extend(['staged'])

    if options.from_status:
        fetch_from = options.from_status.split(',')
        print "Overriding to read from",fetch_from

    for status in fetch_from:
        wfos.extend(session.query(Workflow).filter(Workflow.status==status).all())

    ## in case of partial, go for fetching a list from json ?
    #if options.partial and not specific:
    #    pass

    dataset_endpoints = json.loads(open('%s/dataset_endpoints.json'%monitor_dir).read())
    aaa_mapping = json.loads(open('%s/equalizor.json'%monitor_pub_dir).read())['mapping']

    all_stuck = set()
    all_stuck.update( json.loads( open('%s/stuck_transfers.json'%monitor_dir).read() ))
    all_stuck.update( getAllStuckDataset()) 

    max_per_round = UC.get('max_per_round').get('assignor',None)
    max_cpuh_block = UC.get('max_cpuh_block')
    random.shuffle( wfos )
    for wfo in wfos:
        
        if options.limit and (n_stalled+n_assigned)>options.limit:
            break

        if max_per_round and (n_stalled+n_assigned)>max_per_round:
            break

        if specific:
            if not any(map(lambda sp: sp in wfo.name,specific.split(','))): continue
            #if not specific in wfo.name: continue
        print "\n\n"
        wfh = workflowInfo( url, wfo.name)

        if options.priority and int(wfh.request['RequestPriority']) < options.priority:
            continue

        options_text=""
        if options.early: options_text+=", early option is ON"
        if options.partial: 
            options_text+=", partial option is ON"
            options_text+=", good fraction is %.2f"%options.good_enough
        


        wfh.sendLog('assignor',"%s to be assigned%s"%(wfo.name, options_text))

        ## the site whitelist takes into account siteInfo, campaignInfo, memory and cores
        (lheinput,primary,parent,secondary, sites_allowed) = wfh.getSiteWhiteList()
        output_tiers = list(set([o.split('/')[-1] for o in wfh.request['OutputDatasets']]))
        
        
        is_stuck = (all_stuck & primary)
        if is_stuck:
            wfh.sendLog('assignor',"%s are stuck input"%(','.join( is_stuck)))

        ## check if by configuration we gave it a GO
        no_go = False
        if not wfh.go(log=True) and not options.go:
            no_go = True

        allowed_secondary = {}
        assign_parameters = {}
        check_secondary = False
        for campaign in wfh.getCampaigns():
            if campaign in CI.campaigns:
                assign_parameters.update( CI.campaigns[campaign] )

            if campaign in CI.campaigns and 'secondaries' in CI.campaigns[campaign]:
                if CI.campaigns[campaign]['secondaries']:
                    allowed_secondary.update( CI.campaigns[campaign]['secondaries'] )
                    check_secondary = True
            if campaign in CI.campaigns and 'banned_tier' in CI.campaigns[campaign]:
#.........这里部分代码省略.........

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 stagor(url,specific =None, options=None):
    
    if not componentInfo().check(): return
    SI = siteInfo()
    CI = campaignInfo()
    UC = unifiedConfiguration()

    done_by_wf_id = {}
    done_by_input = {}
    completion_by_input = {}
    good_enough = 100.0
    
    lost = json.loads(open('lost_blocks_datasets.json').read())
    still_lost = []
    for dataset in lost:
        l = findLostBlocks(url ,dataset)
        if not l:
            print dataset,"is not really lost"
        else:
            still_lost.append( dataset )
    open('lost_blocks_datasets.json','w').write( json.dumps( still_lost, indent=2) )

    if options.fast:
        print "doing the fast check of staged with threshold:",options.goodavailability
        for wfo in session.query(Workflow).filter(Workflow.status == 'staging').all():
            if specific and not specific in wfo.name: continue
            wfi = workflowInfo(url, wfo.name)
            sites_allowed = getSiteWhiteList( wfi.getIO() )
            if 'SiteWhitelist' in CI.parameters(wfi.request['Campaign']):
                sites_allowed = CI.parameters(wfi.request['Campaign'])['SiteWhitelist']
            if 'SiteBlacklist' in CI.parameters(wfi.request['Campaign']):
                sites_allowed = list(set(sites_allowed) - set(CI.parameters(wfi.request['Campaign'])['SiteBlacklist']))
            _,primaries,_,secondaries = wfi.getIO()
            se_allowed = [SI.CE_to_SE(site) for site in sites_allowed] 
            all_check = True
            for dataset in list(primaries):#+list(secondaries) ?
                #print se_allowed
                available = getDatasetBlocksFraction( url , dataset , sites=se_allowed )
                all_check &= (available >= options.goodavailability)
                if not all_check: break

            if all_check:
                print "\t\t",wfo.name,"can go staged"
                wfo.status = 'staged'
                session.commit()
            else:
                print "\t",wfo.name,"can wait a bit more"
        return 

    for wfo in session.query(Workflow).filter(Workflow.status == 'staging').all():
        wfi = workflowInfo(url, wfo.name)
        _,primaries,_,secondaries = wfi.getIO()
        for dataset in list(primaries)+list(secondaries):
            done_by_input[dataset] = {}
            completion_by_input[dataset] = {}
            print wfo.name,"needs",dataset

    for transfer in session.query(Transfer).all():
        if specific  and str(transfer.phedexid)!=str(specific): continue

        skip=True
        for wfid in transfer.workflows_id:
            tr_wf = session.query(Workflow).get(wfid)
            if tr_wf: 
                if tr_wf.status == 'staging':
                    print "\t",transfer.phedexid,"is staging for",tr_wf.name
                    skip=False

        if skip: continue
        if transfer.phedexid<0: continue

        ## check the status of transfers
        checks = checkTransferApproval(url,  transfer.phedexid)
        approved = all(checks.values())
        if not approved:
            print transfer.phedexid,"is not yet approved"
            approveSubscription(url, transfer.phedexid)
            continue

        ## check on transfer completion
        checks = checkTransferStatus(url, transfer.phedexid, nocollapse=True)

        if not specific:
            for dsname in checks:
                if not dsname in done_by_input: done_by_input[dsname]={}
                if not dsname in completion_by_input: completion_by_input[dsname] = {}
                done_by_input[dsname][transfer.phedexid]=all(map(lambda i:i>=good_enough, checks[dsname].values()))
                completion_by_input[dsname][transfer.phedexid]=checks[dsname].values()
        if checks:
            print "Checks for",transfer.phedexid,[node.values() for node in checks.values()]
            done = all(map(lambda i:i>=good_enough,list(itertools.chain.from_iterable([node.values() for node in checks.values()]))))
        else:
            ## it is empty, is that a sign that all is done and away ?
            print "ERROR with the scubscriptions API of ",transfer.phedexid
            print "Most likely something else is overiding the transfer request. Need to work on finding the replacement automatically, if the replacement exists"
            done = False

        ## the thing above is NOT giving the right number
        #done = False

#.........这里部分代码省略.........

def getPileupDataset(url, workflow):
	return workflowInfo( url,workflow).getPileupDataset()

#.........这里部分代码省略.........
        sleep_time = min(max(0.5, total_running_time / len(wfs)), 10)

    random.shuffle( wfs )

    in_manual = 0

    ## now you have a record of what file was invalidated globally from TT
    TMDB_invalid = dataCache.get('file_invalidation') 
    #try:
    #    TMDB_invalid = set([row[3] for row in csv.reader( os.popen('curl -s "https://docs.google.com/spreadsheets/d/11fFsDOTLTtRcI4Q3gXw0GNj4ZS8IoXMoQDC3CbOo_2o/export?format=csv"'))])
    #    TMDB_invalid = map(lambda e : e.split(':')[-1], TMDB_invalid)
    #    print len(TMDB_invalid),"globally invalidated files"
    #except Exception as e:
    #    print "TMDB not fetched"
    #    print str(e)
    #    TMDB_invalid = []


    print len(wfs),"to consider, pausing for",sleep_time
    max_per_round = UC.get('max_per_round').get('checkor',None)
    if options.limit: max_per_round=options.limit
    if max_per_round and not spec: wfs = wfs[:max_per_round]



    for wfo in wfs:
        if spec and not (spec in wfo.name): continue
        
        time.sleep( sleep_time )
        
        time_point("Starting with %s"% wfo.name)

        ## get info
        wfi = workflowInfo(url, wfo.name)
        wfi.sendLog('checkor',"checking on %s %s"%( wfo.name,wfo.status))
        ## make sure the wm status is up to date.
        # and send things back/forward if necessary.
        wfo.wm_status = wfi.request['RequestStatus']
        if wfo.wm_status == 'closed-out':
            ## manually closed-out
            wfi.sendLog('checkor',"%s is already %s, setting close"%( wfo.name , wfo.wm_status))
            wfo.status = 'close'
            session.commit()
            continue

        elif wfo.wm_status in ['failed','aborted','aborted-archived','rejected','rejected-archived','aborted-completed']:
            ## went into trouble
            wfo.status = 'trouble'
            wfi.sendLog('checkor',"%s is in trouble %s"%(wfo.name, wfo.wm_status))
            session.commit()
            continue
        elif wfo.wm_status in ['assigned','acquired']:
            ## not worth checking yet
            wfi.sendLog('checkor',"%s is not running yet"%wfo.name)
            session.commit()
            continue
        
        if '-onhold' in wfo.status:
            if wfo.name in holdings and wfo.name not in bypasses:
                wfi.sendLog('checkor',"%s is on hold"%wfo.name)
                continue

        if wfo.wm_status != 'completed': #and not wfo.name in bypasses:
            ## for sure move on with closeout check if in completed
            wfi.sendLog('checkor',"no need to check on %s in status %s"%(wfo.name, wfo.wm_status))
            session.commit()

from assignSession import *
import time
from utils import getWorkLoad, checkTransferStatus, workflowInfo
import pprint
import sys

url = 'cmsweb.cern.ch'

#wfi = workflowInfo( url, "pdmvserv_HIG-2019GEMUpg14DR-00116_00086_v0__150330_112405_8526")
wfi = workflowInfo( url, "pdmvserv_task_B2G-RunIIWinter15wmLHE-00001__v1_T_150402_161327_2265")
print wfi.getLumiWhiteList()

#print wfi.acquisitionEra()

#pid = sys.argv[1]
#tr = session.query(Transfer).filter(Transfer.phedexid== pid).first()
#for wfid in tr.workflows_id:
#    wf = session.query(Workflow).get(wfid)
#    print wf.id,wf.name
#    wf.status = 'staging'
#session.commit()

#checks = checkTransferStatus( 'cmsweb.cern.ch',440053 )
#print checks

#tr = session.query(Transfer).filter(Transfer.phedexid== 440100 ).first()
#session.delete( tr )
#session.commit()

#wf = session.query(Workflow).filter(Workflow.name=="pdmvserv_TOP-Summer12DR53X-00302_00379_v0__150331_100923_4420"
#                                    ).first()