def _upload_job_files_to_hdfs(self, where, job_dir, job, configs, proxy_configs=None):
mains = job.mains or []
libs = job.libs or []
builtin_libs = edp.get_builtin_binaries(job, configs)
uploaded_paths = []
hdfs_user = self.get_hdfs_user()
job_dir_suffix = "lib" if job.type != edp.JOB_TYPE_SHELL else ""
lib_dir = os.path.join(job_dir, job_dir_suffix)
with remote.get_remote(where) as r:
for main in mains:
raw_data = dispatch.get_raw_binary(main, proxy_configs=proxy_configs, remote=r)
if isinstance(raw_data, dict) and raw_data["type"] == "path":
h.copy_from_local(r, raw_data["path"], job_dir, hdfs_user)
else:
h.put_file_to_hdfs(r, raw_data, main.name, job_dir, hdfs_user)
uploaded_paths.append(job_dir + "/" + main.name)
if len(libs) and job_dir_suffix:
# HDFS 2.2.0 fails to put file if the lib dir does not exist
self.create_hdfs_dir(r, lib_dir)
for lib in libs:
raw_data = dispatch.get_raw_binary(lib, proxy_configs=proxy_configs, remote=remote)
if isinstance(raw_data, dict) and raw_data["type"] == "path":
h.copy_from_local(r, raw_data["path"], lib_dir, hdfs_user)
else:
h.put_file_to_hdfs(r, raw_data, lib.name, lib_dir, hdfs_user)
uploaded_paths.append(lib_dir + "/" + lib.name)
for lib in builtin_libs:
h.put_file_to_hdfs(r, lib["raw"], lib["name"], lib_dir, hdfs_user)
uploaded_paths.append(lib_dir + "/" + lib["name"])
return uploaded_paths
def _upload_job_binaries(self, r, job_binaries, proxy_configs,
hdfs_user, job_dir):
uploaded_paths = []
for jb in job_binaries:
path = jb_manager.JOB_BINARIES. \
get_job_binary_by_url(jb.url). \
copy_binary_to_cluster(jb, proxy_configs=proxy_configs,
remote=r, context=context.ctx())
h.copy_from_local(r, path, job_dir, hdfs_user)
uploaded_paths.append(path)
return uploaded_paths
def _upload_job_files_to_hdfs(self,
where,
job_dir,
job,
configs,
proxy_configs=None):
mains = job.mains or []
libs = job.libs or []
builtin_libs = edp.get_builtin_binaries(job, configs)
uploaded_paths = []
hdfs_user = self.get_hdfs_user()
job_dir_suffix = 'lib' if job.type != edp.JOB_TYPE_SHELL else ''
lib_dir = os.path.join(job_dir, job_dir_suffix)
with remote.get_remote(where) as r:
for main in mains:
raw_data = dispatch.get_raw_binary(main,
proxy_configs=proxy_configs,
remote=r)
if isinstance(raw_data, dict) and raw_data["type"] == "path":
h.copy_from_local(r, raw_data['path'], job_dir, hdfs_user)
else:
h.put_file_to_hdfs(r, raw_data, main.name, job_dir,
hdfs_user)
uploaded_paths.append(job_dir + '/' + main.name)
if len(libs) and job_dir_suffix:
# HDFS 2.2.0 fails to put file if the lib dir does not exist
self.create_hdfs_dir(r, lib_dir)
for lib in libs:
raw_data = dispatch.get_raw_binary(lib,
proxy_configs=proxy_configs,
remote=remote)
if isinstance(raw_data, dict) and raw_data["type"] == "path":
h.copy_from_local(r, raw_data['path'], lib_dir, hdfs_user)
else:
h.put_file_to_hdfs(r, raw_data, lib.name, lib_dir,
hdfs_user)
uploaded_paths.append(lib_dir + '/' + lib.name)
for lib in builtin_libs:
h.put_file_to_hdfs(r, lib['raw'], lib['name'], lib_dir,
hdfs_user)
uploaded_paths.append(lib_dir + '/' + lib['name'])
return uploaded_paths
def test_copy_from_local(self):
helper.copy_from_local(self.cluster, 'Galaxy', 'Earth', 'BigBang')
self.cluster.execute_command.assert_called_once_with(
'sudo su - -c "hadoop dfs -copyFromLocal Galaxy Earth" BigBang')
def test_copy_from_local(self):
helper.copy_from_local(self.cluster, 'Galaxy', 'Earth', 'BigBang')
self.cluster.execute_command.assert_called_once_with(
'sudo su - -c "hadoop dfs -copyFromLocal Galaxy Earth" BigBang')
