[hailctl] Add hailctl batch submit

build.yaml

@@ -2642,6 +2642,59 @@ steps:
       - merge_code
       - service_base_image
       - deploy_batch
+  - kind: runImage
+    name: test_hailctl_batch
+    numSplits: 5
+    image:
+      valueFrom: hailgenetics_hail_image.image
+    script: |
+      set -ex
+
+      export HAIL_GENETICS_HAIL_IMAGE="{{ hailgenetics_hail_image.image }}"
+
+      mkdir -p foo
+      echo "bar" > foo/baz.txt
+
+      cat >simple_hail.py <<EOF
+      import hail as hl
+
+      with open('foo/baz.txt') as f:
+          print(f.read())
+
+      hl.utils.range_table(10).collect()
+      EOF
+
+      BATCH_ID=$(hailctl batch submit simple_hail.py --name=test-hailctl-batch-submit --files=foo -o json | jq '.id')
+      STATUS=$(hailctl batch wait --quiet $BATCH_ID | jq -jr '.state')
+      if [ "$STATUS" == "success" ]; then
+          exit 0;
+      else
+          exit 1;
+      fi
+    secrets:
+      - name: worker-deploy-config
+        namespace:
+          valueFrom: default_ns.name
+        mountPath: /deploy-config
+      - name: test-tokens
+        namespace:
+          valueFrom: default_ns.name
+        mountPath: /user-tokens
+      - name: test-gsa-key
+        namespace:
+          valueFrom: default_ns.name
+        mountPath: /test-gsa-key
+      - name: ssl-config-batch-tests
+        namespace:
+          valueFrom: default_ns.name
+        mountPath: /ssl-config
+    dependsOn:
+      - hailgenetics_hail_image
+      - create_deploy_config
+      - create_accounts
+      - default_ns
+      - merge_code
+      - deploy_batch
   - kind: runImage
     name: test_batch_docs
     image:

hail/python/hailtop/aiotools/copy.py

@@ -62,7 +62,8 @@ async def copy(*,
                         transfers,
                         files_listener=make_listener(progress, file_tid),
                         bytes_listener=make_listener(progress, bytes_tid))
-                copy_report.summarize()
+                if verbose:
+                    copy_report.summarize()
 
 
 def make_transfer(json_object: Dict[str, str]) -> Transfer:

hail/python/hailtop/batch_client/aioclient.py

@@ -443,7 +443,7 @@ async def wait(self,
             description += ': '
         if progress is not None:
             return await self._wait(description, progress, disable_progress_bar)
-        with BatchProgressBar() as progress2:
+        with BatchProgressBar(disable=disable_progress_bar) as progress2:
             return await self._wait(description, progress2, disable_progress_bar)
 
     async def debug_info(self):

hail/python/hailtop/hailctl/batch/cli.py

@@ -10,6 +10,7 @@
 from . import log
 from . import job
 from . import billing
+from . import submit
 
 
 def parser():
@@ -39,6 +40,11 @@ def parser():
         help='Delete a batch',
         description='Delete a batch'
     )
+    submit_parser = subparsers.add_parser(
+        'submit',
+        help='Submit a batch',
+        description='Submit a batch',
+    )
     log_parser = subparsers.add_parser(
         'log',
         help='Get log for a job',
@@ -69,6 +75,9 @@ def parser():
     delete_parser.set_defaults(module='delete')
     delete.init_parser(delete_parser)
 
+    submit_parser.set_defaults(module='submit')
+    submit.init_parser(submit_parser)
+
     log_parser.set_defaults(module='log')
     log.init_parser(log_parser)
 
@@ -93,7 +102,8 @@ def main(args):
         'cancel': cancel,
         'log': log,
         'job': job,
-        'wait': wait
+        'wait': wait,
+        'submit': submit,
     }
 
     args, pass_through_args = parser().parse_known_args(args=args)

hail/python/hailtop/hailctl/batch/submit.py

@@ -0,0 +1,71 @@
+import asyncio
+import orjson
+import os
+
+import hailtop.batch as hb
+import hailtop.batch_client.client as bc
+from hailtop import pip_version
+from hailtop.aiotools.copy import copy_from_dict
+from hailtop.config import get_remote_tmpdir, get_user_config_path, get_deploy_config
+from hailtop.utils import unpack_comma_delimited_inputs
+
+HAIL_GENETICS_HAIL_IMAGE = os.environ.get('HAIL_GENETICS_HAIL_IMAGE', f'hailgenetics/hail:{pip_version()}')
+
+
+def init_parser(parser):
+    parser.add_argument('script', type=str, help='Path to script')
+    parser.add_argument('--name', type=str, default='', help='Batch name')
+    parser.add_argument('--image-name', type=str, required=False,
+                        help='Name for Docker image. Defaults to hailgenetics/hail')
+    parser.add_argument('--files', nargs='+', action='append', default=[],
+                        help='Comma-separated list of files or directories to add to the working directory of job')
+    parser.add_argument('-o', type=str, default='text', choices=['text', 'json'])
+
+
+async def async_main(args):
+    script = args.script
+    files = unpack_comma_delimited_inputs(args.files)
+    user_config = get_user_config_path()
+    quiet = args.o != 'text'
+
+    remote_tmpdir = get_remote_tmpdir('hailctl batch submit')
+
+    def cloud_prefix(path):
+        return f'{remote_tmpdir}/{path}'
+
+    b = hb.Batch(name=args.name, backend=hb.ServiceBackend())
+    j = b.new_bash_job()
+    j.image(args.image_name or HAIL_GENETICS_HAIL_IMAGE)
+
+    rel_file_paths = [os.path.relpath(file) for file in files]
+    local_files_to_cloud_files = [{'from': local, 'to': cloud_prefix(local)} for local in rel_file_paths]
+    await copy_from_dict(files=[
+        {'from': script, 'to': cloud_prefix(script)},
+        {'from': str(user_config), 'to': cloud_prefix(user_config)},
+        *local_files_to_cloud_files,
+    ])
+    for file in local_files_to_cloud_files:
+        local_file = file['from']
+        cloud_file = file['to']
+        in_file = b.read_input(cloud_file)
+        j.command(f'ln -s {in_file} {local_file}')
+
+    script_file = b.read_input(cloud_prefix(script))
+    config_file = b.read_input(cloud_prefix(user_config))
+    j.command(f'mkdir -p $HOME/.config/hail && ln -s {config_file} $HOME/.config/hail/config.ini')
+
+    command = 'python3' if script.endswith('.py') else 'bash'
+    j.command(f'{command} {script_file}')
+    batch_handle: bc.Batch = b.run(wait=False, disable_progress_bar=quiet)  # type: ignore
+
+    if args.o == 'text':
+        deploy_config = get_deploy_config()
+        url = deploy_config.external_url('batch', f'/batches/{batch_handle.id}/jobs/1')
+        print(f'Submitted batch {batch_handle.id}, see {url}')
+    else:
+        assert args.o == 'json'
+        print(orjson.dumps({'id': batch_handle.id}).decode('utf-8'))
+
+
+def main(args, pass_through_args, client):  # pylint: disable=unused-argument
+    asyncio.run(async_main(args))

hail/python/hailtop/hailctl/batch/wait.py

@@ -1,9 +1,13 @@
+import json
 import sys
 from .batch_cli_utils import get_batch_if_exists
 
 
 def init_parser(parser):
     parser.add_argument('batch_id', type=int)
+    parser.add_argument("--quiet", "-q",
+                        action="store_true",
+                        help="Do not print a progress bar for the batch")
 
 
 def main(args, pass_through_args, client):  # pylint: disable=unused-argument
@@ -13,4 +17,4 @@ def main(args, pass_through_args, client):  # pylint: disable=unused-argument
         sys.exit(1)
 
     batch = maybe_batch
-    print(batch.wait())
+    print(json.dumps(batch.wait(disable_progress_bar=True)))