added --max-retries to retry failing tasks

DETAILS.md

@@ -89,6 +89,7 @@ We highly recommend to use a default configuration file described in the section
 	cromwell|--cromwell|[cromwell-40.jar](https://github.com/broadinstitute/cromwell/releases/download/40/cromwell-40.jar)|Path or URL for Cromwell JAR file
 	max-concurrent-tasks|--max-concurrent-tasks|1000|Maximum number of concurrent tasks
 	max-concurrent-workflows|--max-concurrent-workflows|40|Maximum number of concurrent workflows
+	max-retries|--max-retries|1|Maximum number of retries for failing tasks
 	disable-call-caching|--disable-call-caching| |Disable Cromwell's call-caching (re-using outputs)
 	backend-file|--backend-file| |Custom Cromwell backend conf file. This will override Caper's built-in backends
 

caper/caper.py

@@ -103,6 +103,7 @@ def __init__(self, args):
         self._disable_call_caching = args.get('disable_call_caching')
         self._max_concurrent_workflows = args.get('max_concurrent_workflows')
         self._max_concurrent_tasks = args.get('max_concurrent_tasks')
+        self._max_retries = args.get('max_retries')
         self._tmp_dir = args.get('tmp_dir')
         self._out_dir = args.get('out_dir')
         if self._out_dir is not None:
@@ -814,6 +815,10 @@ def __create_workflow_opts_json_file(
             template['default_runtime_attributes']['sge_extra_param'] = \
                 self._sge_extra_param
 
+        if self._max_retries is not None:
+            template['default_runtime_attributes']['maxRetries'] = \
+                self._max_retries
+
         # if workflow opts file is given by a user, merge it to template
         if self._workflow_opts is not None:
             f = CaperURI(self._workflow_opts).get_local_file()

caper/caper_args.py

@@ -20,6 +20,7 @@
 DEFAULT_MYSQL_DB_PORT = 3306
 DEFAULT_MAX_CONCURRENT_WORKFLOWS = 40
 DEFAULT_MAX_CONCURRENT_TASKS = 1000
+DEFAULT_MAX_RETRIES = 1
 DEFAULT_PORT = 8000
 DEFAULT_IP = 'localhost'
 DEFAULT_FORMAT = 'id,status,name,str_label,submission'
@@ -230,6 +231,11 @@ def parse_caper_arguments():
         type=int,
         help='Number of concurrent workflows. '
              '"system.max-concurrent-workflows" in backend configuration')
+    group_cromwell.add_argument(
+        '--max-retries', default=DEFAULT_MAX_RETRIES,
+        type=int,
+        help='Number of retries for failing tasks. '
+             'equivalent to "maxRetries" in workflow options JSON file.')
     group_cromwell.add_argument(
         '--disable-call-caching', action='store_true',
         help='Disable Cromwell\'s call caching, which re-uses outputs from '
@@ -527,6 +533,11 @@ def parse_caper_arguments():
             and isinstance(max_concurrent_workflows, str):
         args_d['max_concurrent_workflows'] = int(max_concurrent_workflows)
 
+    max_retries = args_d.get('max_retries')
+    if max_retries is not None \
+            and isinstance(max_retries, str):
+        args_d['max_retries'] = int(max_retries)
+
     # init some important path variables
     if args_d.get('out_dir') is None:
         args_d['out_dir'] = os.getcwd()

setup.py

@@ -5,7 +5,7 @@
 
 setuptools.setup(
     name='caper',
-    version='v0.2.7',
+    version='v0.2.8',
     python_requires='>3.4.1',
     scripts=['bin/caper', 'mysql/run_mysql_server_docker.sh',
              'mysql/run_mysql_server_singularity.sh'],