def upload_timelog( the_root_object, the_study_name, the_booked,
the_upload_seed_size, the_number_threads,
the_item, the_log_file ) :
a_time_log_file = "%s-%s" % ( the_log_file, the_item )
sh_command( "cp %s %s" % ( the_log_file, a_time_log_file ) )
import data_upload; data_upload.entry_point( the_root_object, { a_time_log_file : '' }, the_upload_seed_size, the_number_threads,
the_study_name, the_booked )
pass
def upload_timedata( the_root_object, the_study_name, the_booked,
the_upload_seed_size, the_number_threads,
the_case_dir, the_item ) :
sh_command( "reconstructPar -case '%s' -time %s >> log.reconstructPar 2>&1" % ( the_case_dir, the_item ) )
import os.path; a_timestamp = os.path.join( the_case_dir, the_item )
import os.path; a_timestamp = os.path.abspath( a_timestamp )
import data_upload; data_upload.entry_point( the_root_object, { a_timestamp : '' }, the_upload_seed_size, the_number_threads,
the_study_name, the_booked )
pass
def main() :
#----------------------- Defining utility command-line interface -------------------------
import data_transfer_options
import solver_start_options
import data_seeding_options
import openmpi_config_options
an_usage_description = "%prog"
an_usage_description += ec2.use.options.usage_description()
an_usage_description += data_transfer_options.usage_description()
an_usage_description += solver_start_options.usage_description()
from cloudflu import VERSION
a_version = "%s" % VERSION
from optparse import IndentedHelpFormatter
a_help_formatter = IndentedHelpFormatter( width = 127 )
from optparse import OptionParser
an_option_parser = OptionParser( usage = an_usage_description, version = a_version, formatter = a_help_formatter )
#----------------------- Definition of the command line arguments ------------------------
ec2.use.options.add( an_option_parser )
data_transfer_options.add( an_option_parser )
solver_start_options.add( an_option_parser )
data_seeding_options.add( an_option_parser )
common.concurrency_options.add( an_option_parser )
openmpi_config_options.add( an_option_parser )
amazon.security_options.add( an_option_parser )
common.options.add( an_option_parser )
#------------------ Extracting and verifying command-line arguments ----------------------
an_options, an_args = an_option_parser.parse_args()
an_enable_debug, a_log_file = common.options.extract( an_option_parser )
AWS_ACCESS_KEY_ID, AWS_SECRET_ACCESS_KEY = amazon.security_options.extract( an_option_parser )
a_cluster_id = ec2.use.options.extract( an_option_parser )
a_hostfile = openmpi_config_options.extract( an_option_parser )
a_number_threads = common.concurrency_options.extract( an_option_parser )
an_upload_seed_size = data_seeding_options.extract( an_option_parser )
a_case_dir, a_run_hook = solver_start_options.extract( an_option_parser )
a_study_name = data_transfer_options.extract( an_option_parser )
from cloudflu.preferences import get
a_cluster_location = get( 'amazon.cluster.location' )
a_host_port = int( get( 'amazon.cluster.host_port' ) )
a_login_name = get( 'amazon.cluster.login_name' )
print_d( "\n--------------------------- Canonical substitution ------------------------\n" )
import sys; an_engine = sys.argv[ 0 ]
a_call = "%s %s %s %s %s" % ( an_engine,
ec2.use.options.compose( a_cluster_id ),
data_transfer_options.compose( a_study_name ),
solver_start_options.compose( a_case_dir, a_run_hook ),
openmpi_config_options.compose( a_hostfile ) )
print_d( a_call + '\n' )
print_d( "\n----------------------- Running actual functionality ----------------------\n" )
a_spent_time = Timer()
an_ec2_conn = ec2.common.region_connect( a_cluster_location, AWS_ACCESS_KEY_ID, AWS_SECRET_ACCESS_KEY )
a_reservation = ec2.use.get_reservation( an_ec2_conn, a_cluster_id )
print_d( '< %r > : %s\n' % ( a_reservation, a_reservation.instances ) )
a_master_node = an_instance = a_reservation.instances[ 0 ]
print_d( "\n--------------------- Uploading case data to S3 ---------------------------\n" )
import os; a_case_name = os.path.basename( a_case_dir )
from cloudflu.amazon.s3 import TRootObject
a_root_object = TRootObject.get( AWS_ACCESS_KEY_ID, AWS_SECRET_ACCESS_KEY )
print_d( "a_root_object = %s\n" % a_root_object )
import data_upload
an_input_study = data_upload.entry_point( a_root_object, { a_case_dir : '' }, an_upload_seed_size, a_number_threads )
an_input_study_name = an_input_study.name()
print_d( "an_input_study_name = '%s'\n" % an_input_study_name )
print_d( "\n------------------ Installing cloudflu to master node ----------------------\n" )
a_password = None
a_host_name = an_instance.public_dns_name
an_identity_file = ec2.run.get_identity_filepath( an_instance.key_name )
ssh.options.echo( a_password, an_identity_file, a_host_port, a_login_name, a_host_name )
a_ssh_client = ssh.connect( a_password, an_identity_file, a_host_port, a_login_name, a_host_name )
an_instance2ssh = {}
an_instance2ssh[ an_instance ] = a_ssh_client
print_d( "\n------------- Downloading case data from S3 to the master node ------------\n" )
a_stdout_lines = ssh.command( a_ssh_client, 'python -c "import os, os.path, tempfile; print tempfile.mkdtemp()"' )
a_working_dir = a_stdout_lines[ 0 ][ : -1 ]
print_d( "a_working_dir = '%s'\n" % a_working_dir )
ssh.command( a_ssh_client, "cloudflu-download --study-name=%s --output-dir=%s --remove" % ( an_input_study_name, a_working_dir ) )
print_d( "\n--- Sharing the solver case folder for all the cluster nodes through NFS --\n" )
ssh.command( a_ssh_client, "sudo sh -c 'echo %s *\(rw,no_root_squash,sync,subtree_check\) >> /etc/exports'" % ( a_working_dir ) )
ssh.command( a_ssh_client, "sudo exportfs -a" ) # make changes effective on the running NFS server
for an_instance in a_reservation.instances[ 1 : ] :
a_host_name = an_instance.public_dns_name
ssh.options.echo( a_password, an_identity_file, a_host_port, a_login_name, a_host_name )
a_ssh_client = ssh.connect( a_password, an_identity_file, a_host_port, a_login_name, a_host_name )
ssh.command( a_ssh_client, "mkdir -p %s" % ( a_working_dir ) )
ssh.command( a_ssh_client, "sudo mount %s:%s %s" % ( a_master_node.private_ip_address, a_working_dir, a_working_dir ) )
an_instance2ssh[ an_instance ] = a_ssh_client
pass
print_d( "\n------------------------ Get access the booked study ----------------------\n" )
import study_book;
an_output_study = study_book.entry_point( a_root_object, a_study_name, True )
print_d( "\n----------------------- Running of the solver case ------------------------\n" )
a_num_nodes = len( a_reservation.instances )
a_target_dir = os.path.join( a_working_dir, a_case_name )
an_additional_args = "--study-name='%s' --booked --upload-seed-size=%d" % ( a_study_name, an_upload_seed_size )
a_ssh_client = an_instance2ssh[ a_master_node ]
ssh.command( a_ssh_client, "%s/%s %d '%s' %s" % ( a_target_dir, a_run_hook, a_num_nodes, a_hostfile, an_additional_args ) )
an_output_study.seal()
[ a_ssh_client.close() for a_ssh_client in an_instance2ssh.values() ]
print_d( "a_spent_time = %s, sec\n" % a_spent_time )
print_d( "\n------------------ Printing succussive pipeline arguments -----------------\n" )
ec2.use.options.track( a_cluster_id )
print_d( "\n--------------------------- Canonical substitution ------------------------\n" )
print_d( a_call + '\n' )
print_d( "\n-------------------------------------- OK ---------------------------------\n" )
pass
