def _upload_job_files_to_hdfs(self, where, job_dir, job, configs,
proxy_configs=None):
mains = list(job.mains) if job.mains else []
libs = list(job.libs) if job.libs else []
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:
job_binaries = mains + libs
self._prepare_job_binaries(job_binaries, r)
# upload mains
uploaded_paths.extend(self._upload_job_binaries(r, mains,
proxy_configs,
hdfs_user,
job_dir))
# upload libs
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)
uploaded_paths.extend(self._upload_job_binaries(r, libs,
proxy_configs,
hdfs_user,
lib_dir))
# upload buitin_libs
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_put_file_to_hdfs(self, open_get, mock_six):
open_get.return_value = '/tmp/workflow.xml'
mock_six.return_value = 111
helper.put_file_to_hdfs(self.cluster, open_get, 'workflow', '/tmp',
'hdfs')
self.cluster.execute_command.assert_called_once_with(
'sudo su - -c "hadoop dfs -copyFromLocal /tmp/workflow.111'
' /tmp/workflow" hdfs && sudo rm -f /tmp/workflow.111')
def test_put_file_to_hdfs(self, open_get, mock_six):
open_get.return_value = '/tmp/workflow.xml'
mock_six.return_value = 111
helper.put_file_to_hdfs(self.cluster, open_get, 'workflow',
'/tmp', 'hdfs')
self.cluster.execute_command.assert_called_once_with(
'sudo su - -c "hadoop dfs -copyFromLocal /tmp/workflow.111'
' /tmp/workflow" hdfs && sudo rm -f /tmp/workflow.111')
def upload_job_files(where, job_dir, job, hdfs_user):
mains = job.mains or []
libs = job.libs or []
uploaded_paths = []
with remote.get_remote(where) as r:
for main in mains:
raw_data = dispatch.get_raw_binary(main)
h.put_file_to_hdfs(r, raw_data, main.name, job_dir, hdfs_user)
uploaded_paths.append(job_dir + '/' + main.name)
for lib in libs:
raw_data = dispatch.get_raw_binary(lib)
# HDFS 2.2.0 fails to put file if the lib dir does not exist
h.create_dir(r, job_dir + "/lib", hdfs_user)
h.put_file_to_hdfs(r, raw_data, lib.name, job_dir + "/lib",
hdfs_user)
uploaded_paths.append(job_dir + '/lib/' + lib.name)
return uploaded_paths
def upload_job_files(where, job_dir, job, hdfs_user):
mains = job.mains or []
libs = job.libs or []
uploaded_paths = []
with remote.get_remote(where) as r:
for main in mains:
raw_data = dispatch.get_raw_binary(main)
h.put_file_to_hdfs(r, raw_data, main.name, job_dir, hdfs_user)
uploaded_paths.append(job_dir + '/' + main.name)
for lib in libs:
raw_data = dispatch.get_raw_binary(lib)
# HDFS 2.2.0 fails to put file if the lib dir does not exist
h.create_dir(r, job_dir + "/lib", hdfs_user)
h.put_file_to_hdfs(r, raw_data, lib.name, job_dir + "/lib",
hdfs_user)
uploaded_paths.append(job_dir + '/lib/' + lib.name)
return uploaded_paths
def _upload_job_files_to_hdfs(self, where, job_dir, job,
proxy_configs=None):
mains = job.mains or []
libs = job.libs or []
uploaded_paths = []
hdfs_user = self.get_hdfs_user()
with remote.get_remote(where) as r:
for main in mains:
raw_data = dispatch.get_raw_binary(main, proxy_configs)
h.put_file_to_hdfs(r, raw_data, main.name, job_dir, hdfs_user)
uploaded_paths.append(job_dir + '/' + main.name)
for lib in libs:
raw_data = dispatch.get_raw_binary(lib, proxy_configs)
# HDFS 2.2.0 fails to put file if the lib dir does not exist
self.create_hdfs_dir(r, job_dir + "/lib")
h.put_file_to_hdfs(r, raw_data, lib.name, job_dir + "/lib",
hdfs_user)
uploaded_paths.append(job_dir + '/lib/' + lib.name)
return uploaded_paths
def _upload_job_files_to_hdfs(self, where, job_dir, job,
proxy_configs=None):
mains = job.mains or []
libs = job.libs or []
uploaded_paths = []
hdfs_user = self.get_hdfs_user()
with remote.get_remote(where) as r:
for main in mains:
raw_data = dispatch.get_raw_binary(main, proxy_configs)
h.put_file_to_hdfs(r, raw_data, main.name, job_dir, hdfs_user)
uploaded_paths.append(job_dir + '/' + main.name)
for lib in libs:
raw_data = dispatch.get_raw_binary(lib, proxy_configs)
# HDFS 2.2.0 fails to put file if the lib dir does not exist
self.create_hdfs_dir(r, job_dir + "/lib")
h.put_file_to_hdfs(r, raw_data, lib.name, job_dir + "/lib",
hdfs_user)
uploaded_paths.append(job_dir + '/lib/' + lib.name)
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)
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)
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_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()
lib_dir = job_dir + '/lib'
with remote.get_remote(where) as r:
for main in mains:
raw_data = dispatch.get_raw_binary(main, proxy_configs)
h.put_file_to_hdfs(r, raw_data, main.name, job_dir, hdfs_user)
uploaded_paths.append(job_dir + '/' + main.name)
if len(libs) > 0:
# 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)
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_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_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_workflow_file(self, where, job_dir, wf_xml, hdfs_user):
with remote.get_remote(where) as r:
h.put_file_to_hdfs(r, wf_xml, "workflow.xml", job_dir, hdfs_user)
return "%s/workflow.xml" % job_dir
def _upload_workflow_file(self, where, job_dir, wf_xml, hdfs_user):
with remote.get_remote(where) as r:
h.put_file_to_hdfs(r, wf_xml, "workflow.xml", job_dir, hdfs_user)
return "%s/workflow.xml" % job_dir
def _upload_coordinator_file(self, where, job_dir, wf_xml, hdfs_user):
with remote.get_remote(where) as r:
h.put_file_to_hdfs(r, wf_xml, "coordinator.xml", job_dir,
hdfs_user)
return "%s/coordinator.xml" % job_dir
def _upload_coordinator_file(self, where, job_dir, wf_xml, hdfs_user):
with remote.get_remote(where) as r:
h.put_file_to_hdfs(r, wf_xml, "coordinator.xml", job_dir,
hdfs_user)
return "%s/coordinator.xml" % job_dir