def putCurrentClusters(self, clusters):
jobManager = ClusterJobManager()
existingClusters = list(self.getCurrentClusters())
newClusters = [
cluster for cluster in clusters if cluster not in existingClusters
]
expiredClusters = [
cluster for cluster in existingClusters if cluster not in clusters
]
for cluster in newClusters:
job = WorkerJob(JOB_PROCESSNEWCLUSTER,
{JOBARG_PROCESSNEWCLUSTER_CLUSTER: list(cluster)})
jobManager.enqueueJob(job)
logging.info("Put process new cluster job. Cluster id: %s.",
cluster.id)
logging.info("Number of clusters to delete are: %i",
len(expiredClusters))
self.clusterTableManager.deleteClusters(expiredClusters)
def archiveStaleDocs():
"""
Remove the docs fro current working set
Run this job periodically.
"""
clusterManager = ClusterManager()
jobManager = MinerJobManager()
logging.info("Archiving old clusters.")
staleClusters = clusterManager.archiveOldClusters()
for cluster in staleClusters:
for docKey in cluster:
job = WorkerJob(JOB_CLEANUPDOC, { JOBARG_CLEANUPDOC_DOCID : docKey})
jobManager.enqueueJob(job)
logging.info(
"Put cleanup doc job for docId: %s. Job id: %s",
docKey,
job.jobId)
logging.info("Archived old clusters and cleaned up docs in them from working set.")
