def listSlotsForQueue(queue):
output = yield commands.getOutput(['qconf', '-sq', queue], log=True)
conf = _parseSGEConf(output['stdout'])
slots = conf['slots'].split(',')
clusterWide = [s for s in slots if s[0] != '[']
nodeSpecific = [s[1:-1].split('=') for s in slots if s[0] == '[']
defer.returnValue({'cluster': int(clusterWide[0]),
'nodes': dict([(h, int(v)) for h, v in nodeSpecific])})
def listSlotsForQueue(queue):
output = yield commands.getOutput(['qconf', '-sq', queue], log=True)
conf = _parseSGEConf(output['stdout'])
slots = conf['slots'].split(',')
clusterWide = [s for s in slots if s[0] != '[']
nodeSpecific = [s[1:-1].split('=') for s in slots if s[0] == '[']
defer.returnValue({
'cluster': int(clusterWide[0]),
'nodes': dict([(h, int(v)) for h, v in nodeSpecific])
})
def _createSupervisor(state):
log.msg('LOAD: Starting')
output = yield commands.getOutput(['hostname', '-f'], log=True)
state.hostname = output['stdout'].strip()
# Let's get our queue information
execSlots = yield sge_queue.listSlotsForQueue(EXEC_QUEUE)
stagingSlots = yield sge_queue.listSlotsForQueue(STAGING_QUEUE)
state.master = MachineInformation()
state.master.setExecSlots(execSlots['nodes'].get(state.hostname, execSlots['cluster']))
state.master.setStagingSlots(stagingSlots['nodes'].get(state.hostname, stagingSlots['cluster']))
reactor.callLater(0.0, _loopSupervisor, state)
def _createSupervisor(state):
log.msg('LOAD: Starting')
output = yield commands.getOutput(['hostname', '-f'], log=True)
state.hostname = output['stdout'].strip()
# Let's get our queue information
execSlots = yield sge_queue.listSlotsForQueue(EXEC_QUEUE)
stagingSlots = yield sge_queue.listSlotsForQueue(STAGING_QUEUE)
state.master = MachineInformation()
state.master.setExecSlots(execSlots['nodes'].get(state.hostname,
execSlots['cluster']))
state.master.setStagingSlots(stagingSlots['nodes'].get(
state.hostname, stagingSlots['cluster']))
reactor.callLater(0.0, _loopSupervisor, state)
def getOutput(host, cmd, sshUser=None, sshFlags=None, initialText=None, log=False, **kwargs):
command = ['ssh']
if sshUser:
host = sshUser + '@' + host
if sshFlags:
command.append(sshFlags)
command.append(host)
command.append(core.quoteEscape(cmd))
command = ' '.join(command)
return commands.getOutput(commands.shell(str(command)),
initialText=str(initialText),
log=log,
**kwargs)
def _getOutput(batchState, command, *args, **kwargs):
_log(batchState, 'CMD: ' + ' '.join(command))
return commands.getOutput(command, *args, **kwargs)
clusters = yield clusters_www.listClusters('localhost',
{'cluster_name': clusterName},
'guest')
cluster = clusters[0]
yield pipeline_www.resumePipeline(cluster['master']['public_dns'],
'local',
pipeline.userName,
child)
except Exception, err:
log.err('Error trying to resume child')
log.err(err)
# Reset the pipeline task to IDLE
yield tasks_tx.updateTask(pipeline.taskName,
lambda t : t.setState(tasks_tx.task.TASK_IDLE))
if pipeline.pipelineId:
cmd = ['resume_pipeline.pl',
'--pipeline_id=' + pipeline.pipelineId,
'--taskname=' + pipeline.taskName]
if pipeline.queue:
cmd.append('--queue=' + pipeline.queue)
yield commands.getOutput(cmd, log=True)
defer.returnValue(pipeline)
def _getOutput(batchState, command, *args, **kwargs):
_log(batchState, 'CMD: ' + ' '.join(command))
return commands.getOutput(command, *args, **kwargs)
