def post_start(self, cluster_context, instances):
sentry_host = cluster_context.get_instance(SENTRY)
source = self.conf_dir(cluster_context) + '/global-policy.ini'
with sentry_host.remote() as r:
mfs.mkdir(r, '/user/mapr/sentry', run_as='mapr')
mfs.chmod(r, '/user/mapr/sentry', 777, run_as='mapr')
mfs.copy_from_local(r, source, self.GLOBAL_POLICY_FILE,
hdfs_user='******')
def _copy_jar_files_to_mfs(self, cluster_context):
hive_service = self._hive(cluster_context)
paths = [self._assembly_jar_path(cluster_context)]
if hive_service:
hive_conf = self._hive(cluster_context).conf_dir(cluster_context)
paths.append('%s/hive-site.xml' % hive_conf)
paths += self._hive_datanucleus_libs_paths(cluster_context)
target = self.JAR_FILE_TARGET
hdfs_user = '******'
with cluster_context.get_instance(SPARK_HISTORY_SERVER).remote() as r:
for path in paths:
mfs.copy_from_local(r, path, target, hdfs_user)
def _copy_jar_files_to_mfs(self, cluster_context):
hive_service = self._hive(cluster_context)
paths = [self._assembly_jar_path(cluster_context)]
if hive_service:
hive_conf = self._hive(cluster_context).conf_dir(cluster_context)
paths.append('%s/hive-site.xml' % hive_conf)
paths += self._hive_datanucleus_libs_paths(cluster_context)
target = self.JAR_FILE_TARGET
hdfs_user = '******'
with cluster_context.get_instance(SPARK_HISTORY_SERVER).remote() as r:
for path in paths:
mfs.copy_from_local(r, path, target, hdfs_user)
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 where.remote() as r:
for m in mains:
path = jb_manager.JOB_BINARIES. \
get_job_binary_by_url(m.url). \
copy_binary_to_cluster(m, proxy_configs=proxy_configs,
remote=r, context=context.ctx())
target = os.path.join(job_dir, m.name)
mfs.copy_from_local(r, path, target, hdfs_user)
uploaded_paths.append(target)
if len(libs) > 0:
self.create_hdfs_dir(r, lib_dir)
for l in libs:
path = jb_manager.JOB_BINARIES. \
get_job_binary_by_url(l.url). \
copy_binary_to_cluster(l, proxy_configs=proxy_configs,
remote=r, context=context.ctx())
target = os.path.join(lib_dir, l.name)
mfs.copy_from_local(r, path, target, hdfs_user)
uploaded_paths.append(target)
for lib in builtin_libs:
mfs.put_file_to_maprfs(r, lib['raw'], lib['name'], lib_dir,
hdfs_user)
uploaded_paths.append(lib_dir + '/' + lib['name'])
return uploaded_paths
def copy_assembly_jar_to_mfs(self, cluster_context):
target = self.JAR_FILE_TARGET
hdfs_user = '******'
with cluster_context.get_instance(SPARK_HISTORY_SERVER).remote() as r:
mfs.copy_from_local(r, self._assembly_jar_path(cluster_context),
target, hdfs_user)