13976: Add code comment about only_real in upload_dependencies

[arvados.git] / sdk / cwl / arvados_cwl / runner.py

diff --git a/sdk/cwl/arvados_cwl/runner.py b/sdk/cwl/arvados_cwl/runner.py
index 87acf15d657c89dec9221bf3a6b2fb1d3f612e32..29c0535d93c8a0ff164af01949ae152f056591d7 100644 (file)
--- a/sdk/cwl/arvados_cwl/runner.py
+++ b/sdk/cwl/arvados_cwl/runner.py
@@ -7,7 +7,7 @@ import urlparse
 from functools import partial
 import logging
 import json
-import subprocess
+import subprocess32 as subprocess
 from collections import namedtuple
 
 from StringIO import StringIO
@@ -129,6 +129,8 @@ def upload_dependencies(arvrunner, name, document_loader,
 
     sc = []
     def only_real(obj):
+        # Only interested in local files than need to be uploaded,
+        # don't include file literals, keep references, etc.
         if obj.get("location", "").startswith("file:"):
             sc.append(obj)
 
@@ -168,8 +170,13 @@ def upload_dependencies(arvrunner, name, document_loader,
 
     visit_class(workflowobj, ("CommandLineTool", "Workflow"), discover_default_secondary_files)
 
-    for d in discovered:
-        sc.extend(discovered[d])
+    for d in list(discovered.keys()):
+        # Only interested in discovered secondaryFiles which are local
+        # files that need to be uploaded.
+        if d.startswith("file:"):
+            sc.extend(discovered[d])
+        else:
+            del discovered[d]
 
     mapper = ArvPathMapper(arvrunner, sc, "",
                            "keep:%s",
@@ -353,7 +360,7 @@ class Runner(object):
     def __init__(self, runner, tool, job_order, enable_reuse,
                  output_name, output_tags, submit_runner_ram=0,
                  name=None, on_error=None, submit_runner_image=None,
-                 intermediate_output_ttl=0, merged_map=None, default_storage_classes="default",
+                 intermediate_output_ttl=0, merged_map=None,
                  priority=None, secret_store=None):
         self.arvrunner = runner
         self.tool = tool
@@ -376,15 +383,26 @@ class Runner(object):
         self.intermediate_output_ttl = intermediate_output_ttl
         self.priority = priority
         self.secret_store = secret_store
-        self.default_storage_classes = default_storage_classes
+
+        self.submit_runner_cores = 1
+        self.submit_runner_ram = 1024  # defaut 1 GiB
+
+        runner_resource_req, _ = self.tool.get_requirement("http://arvados.org/cwl#WorkflowRunnerResources")
+        if runner_resource_req:
+            if runner_resource_req.get("coresMin"):
+                self.submit_runner_cores = runner_resource_req["coresMin"]
+            if runner_resource_req.get("ramMin"):
+                self.submit_runner_ram = runner_resource_req["ramMin"]
 
         if submit_runner_ram:
+            # Command line / initializer overrides default and/or spec from workflow
             self.submit_runner_ram = submit_runner_ram
-        else:
-            self.submit_runner_ram = 3000
 
         if self.submit_runner_ram <= 0:
-            raise Exception("Value of --submit-runner-ram must be greater than zero")
+            raise Exception("Value of submit-runner-ram must be greater than zero")
+
+        if self.submit_runner_cores <= 0:
+            raise Exception("Value of submit-runner-cores must be greater than zero")
 
         self.merged_map = merged_map or {}
 
@@ -415,7 +433,7 @@ class Runner(object):
                                                            api_client=self.arvrunner.api,
                                                            keep_client=self.arvrunner.keep_client,
                                                            num_retries=self.arvrunner.num_retries)
-                done.logtail(logc, logger, "%s error log:" % self.arvrunner.label(self), maxlen=40)
+                done.logtail(logc, logger.error, "%s (%s) error log:" % (self.arvrunner.label(self), record["uuid"]), maxlen=40)
 
             self.final_output = record["output"]
             outc = arvados.collection.CollectionReader(self.final_output,