def s3cmdGET(s3cfg_path, s3_path, local_path, shellOutput=True):
if s3cmdLS(s3cfg_path, s3_path, shellOutput):
retries = 3
sleep = 5
while retries > 0:
retries -= 1
cmd = "s3cmd -c %s get --force %s %s" %(s3cfg_path, s3_path, local_path)
output = auxiliary.execute(cmd, shellOutput=shellOutput)
if output.strip()[-4:] == 'done':
return True
time.sleep(sleep)
return False
def s3cmdLS(s3cfg_path, s3_path, shellOutput=True):
cmd = "s3cmd -c %s ls %s" %(s3cfg_path, s3_path)
if auxiliary.execute(cmd, shellOutput=shellOutput):
return True
else:
return False
def s3cmdDEL(s3cfg_path, s3_path, shellOutput=True):
cmd = "s3cmd -c %s del %s" %(s3cfg_path, s3_path)
auxiliary.execute(cmd, shellOutput=shellOutput)
def s3cmdPUT(s3cfg_path, local_path, s3_path, shellOutput=True):
cmd = "s3cmd -c %s put %s %s" %(s3cfg_path, local_path, s3_path)
auxiliary.execute(cmd, shellOutput=shellOutput)
def main(config):
ts_start = datetime.now()
logger = auxiliary.getLogger()
s3cfg = auxiliary.read_option(config, 'general', 's3cfg')
app_input_s3_path_full = auxiliary.read_option(config, 'general', 'app_input_file')
app_output_s3_path_full = auxiliary.read_option(config, 'general', 'app_output_file')
licence = auxiliary.read_option(config, 'general', 'licence')
grok = auxiliary.read_option(config, 'general', 'grok')
hgs = auxiliary.read_option(config, 'general', 'hgs')
current_dir = os.getcwd()
app_input_name = os.path.basename(app_input_s3_path_full)
app_input_path = os.path.join(current_dir, app_input_name)
app_output_name = os.path.basename(app_output_s3_path_full)
app_output_path = os.path.join(current_dir, app_output_name)
logger.info('Starting %s script...\n' % (__name__))
logger.info('Configuration options:')
logger.info('Current directory: %s'% (current_dir))
logger.info('Input archive location: %s'% (app_input_s3_path_full))
logger.info('Output archive location: %s'% (app_output_s3_path_full))
logger.info('GROK location: %s'% (grok))
logger.info('HGS location: %s\n'% (hgs))
logger.info('Downloading input archive')
s3tools.s3cmdGET(s3cfg, app_input_s3_path_full, app_input_path, shellOutput=False)
hgs_file = auxiliary.seekZipForElement(app_input_path, AppTypes.HGS)[0]
working_dir = os.path.join(current_dir, os.path.split(hgs_file)[0])
root_dir = os.path.join(current_dir, auxiliary.getRootDir(hgs_file))
logger.info('Working directory: %s'% (working_dir))
logger.info('Extracting input archive...')
auxiliary.extractZip(app_input_path, current_dir)
os.chdir(working_dir)
cmd = '%s 1>&1 | tee -a %s' %(grok, 'grok.log')
logger.info('Executing GROK: %s' % (cmd))
auxiliary.execute(cmd)
cmd = '%s 1>&1 | tee -a %s' %(hgs, 'hgs.log')
logger.info('Executing HGS: %s' % (cmd))
auxiliary.execute(cmd)
os.chdir(current_dir)
logger.info('Creating output archive...')
auxiliary.createZip(app_output_path[:-4], root_dir)
logger.info('Uploading output archive')
#while not s3tools.s3cmdLS(s3cfg, app_output_s3_path_full, shellOutput=False):
s3tools.s3cmdPUT(s3cfg, app_output_path, app_output_s3_path_full, shellOutput=False)
logger.info('Done.')
ts_end = datetime.now()
ts_dif = ts_end - ts_start
print 'GROK & HGS execution + claud I/O operations took: %s seconds' % ts_dif.total_seconds()
print 'Executed on machine: %s' % messaging.getIPAdreess()
