def start(self, basedir):
LOG.info("Starting CDSW runner...")
repo_type_env = OsUtils.get_env_value(
BranchComparatorEnvVar.REPO_TYPE.value, RepoType.DOWNSTREAM.value)
repo_type: RepoType = RepoType[repo_type_env.upper()]
if repo_type == RepoType.DOWNSTREAM:
self.run_clone_downstream_repos_script(basedir)
elif repo_type == RepoType.UPSTREAM:
# If we are in upstream mode, make sure downstream dir exist
# Currently, yarndevtools requires both repos to be present when initializing.
# BranchComparator is happy with one single repository, upstream or downstream, exclusively.
# Git init the other repository so everything will be alright
FileUtils.create_new_dir(HADOOP_CLOUDERA_BASEDIR)
FileUtils.change_cwd(HADOOP_CLOUDERA_BASEDIR)
os.system("git init")
self.run_clone_upstream_repos_script(basedir)
# TODO investigate why legacy script fails!
self.run_comparator_and_send_mail(repo_type,
algorithm="simple",
run_legacy_script=False)
self.run_comparator_and_send_mail(repo_type,
algorithm="grouped",
run_legacy_script=False)
def run(cls,
command,
working_dir=None,
log_stdout=False,
log_stderr=False,
log_command_result=False,
fail_on_error=False,
fail_message="",
wait_after=0,
wait_message=""):
if working_dir:
FileUtils.change_cwd(working_dir)
args = shlex.split(command)
proc = subprocess.run(args,
universal_newlines=True,
stdout=subprocess.PIPE,
stderr=subprocess.PIPE)
args2 = str(proc.args)
if working_dir:
FileUtils.reset_cwd()
command_result = RegularCommandResult(command, args2, proc.stdout,
proc.stderr, proc.returncode)
if log_command_result:
LOG.info("Command result: %s", command_result)
if log_stdout:
LOG.info("Stdout:\n %s", command_result.stdout)
if log_stderr:
LOG.info("Stderr:\n %s", command_result.stderr)
if fail_on_error and command_result.exit_code != 0:
if fail_message:
LOG.error(fail_message)
raise ValueError(
"Execution failed! Command was: {}".format(command))
if wait_after > 0:
LOG.info("Waiting for %d seconds %s", wait_message)
time.sleep(wait_after)