def getCluster(self):
# get the error counts for the clusters
#clusterErrors=gJobStatistics.getErrorCounts()
clusterErrors = gJobStatistics.getRelativeErrorCounts()
# get the statistics for the clusters
clusterStats = gJobStatistics.getClusterMap()
# use the current set of known clusters as the candidate for excluding
clusterNames = clusterStats.keys()
worstCluster = None
worstWaiting = 0
for cluster in clusterNames:
if clusterStats.has_key(cluster):
nWaiting = clusterStats[cluster].getJobsWaitingToRun()
else:
nWaiting = 0
# Here is the greylisting part: we simply count each error as a
# waiting job so the more errors there are the more unlikely it is
# that a job will end up in the cluster
if clusterErrors.has_key(cluster):
nWaiting = nWaiting + clusterErrors[cluster]
if nWaiting > worstWaiting:
worstWaiting = nWaiting
worstCluster = cluster
if worstCluster != None:
return '-%s' % worstCluster
else:
return None
def getCluster(self):
# get the error counts for the clusters
#clusterErrors=gJobStatistics.getErrorCounts()
clusterErrors = gJobStatistics.getRelativeErrorCounts()
# get the statistics for the clusters
clusterStats = gJobStatistics.getClusterMap()
# get the list of configured resources
clusterNames = [
item.strip() for item in gConfig.get('clusters', '').split(",")
]
bestCluster = None
bestWaiting = 9999999
for cluster in clusterNames:
if clusterStats.has_key(cluster):
nWaiting = clusterStats[cluster].getJobsWaitingToRun()
else:
nWaiting = 0
# Here is the graylisting part: we simply count each error as a
# waiting job so the more errors there are the more unlikely it is
# that a job will end up in the cluster
if clusterErrors.has_key(cluster):
nWaiting = nWaiting + clusterErrors[cluster]
if nWaiting < bestWaiting:
bestWaiting = nWaiting
bestCluster = cluster
return bestCluster
def getCluster(self):
clusterStats = gJobStatistics.getClusterMap()
clusterNames = clusterStats.keys()
worstCluster = None
worstWaiting = 0
for cluster in clusterNames:
if clusterStats[cluster].getJobsWaitingToRun() > worstWaiting:
worstWaiting = clusterStats[cluster].getJobsWaitingToRun()
worstCluster = cluster
if worstCluster != None:
return '-%s' % worstCluster
else:
return None
def getCluster(self):
# get the error counts for the clusters
#clusterErrors=gJobStatistics.getErrorCounts()
clusterErrors = gJobStatistics.getRelativeErrorCounts()
# get the statistics for the clusters
clusterStats = gJobStatistics.getClusterMap()
# use the current set of clusters as the candidate for excluding
clusterNames = clusterStats.keys()
worstCluster = None
worstWaiting = 0
for cluster in clusterNames:
if clusterStats.has_key(cluster):
nWaiting = clusterStats[cluster].getJobsWaitingToRun()
else:
nWaiting = 0
# Here is the blacklisting part: we simply count each error as a
# waiting job so the more errors there are the more unlikely it is
# that a job will end up in the cluster
if clusterErrors.has_key(cluster):
nWaiting = nWaiting + clusterErrors[cluster]
if nWaiting > worstWaiting:
worstWaiting = nWaiting
worstCluster = cluster
clusterNames = [
item.strip() for item in gConfig.get('clusters', '').split(",")
]
if '' in clusterNames:
clusterNames.remove('')
if worstCluster != None:
# We have the worst cluster pinpointed, construct a list of all the others and
# let ARC do the selection
debug('ExcludingARCBroker: excluding %s' % worstCluster)
if worstCluster in clusterNames:
clusterNames.remove(worstCluster)
if len(clusterNames) < 1:
return None
res = ""
for name in clusterNames:
res = "%s -c %s" % (res, name)
return res.strip()
def getCluster(self):
# get the statistics for the clusters
clusterStats = gJobStatistics.getClusterMap()
# get the list of configured resources
clusterNames = [
item.strip() for item in gConfig.get('clusters', '').split(",")
]
bestCluster = None
bestWaiting = 9999999
for cluster in clusterNames:
if clusterStats.has_key(cluster):
nWaiting = clusterStats[cluster].getJobsWaitingToRun()
else:
nWaiting = 0
if nWaiting < bestWaiting:
bestWaiting = nWaiting
bestCluster = cluster
return bestCluster