18842: Initial work adding disk cache option to configuration

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

diff --git a/sdk/cwl/arvados_cwl/arvcontainer.py b/sdk/cwl/arvados_cwl/arvcontainer.py
index f3e122e603f8cd9cf9b768158132f28d7190b685..9b5f322338275fee56035d03909ee5bf23819b20 100644 (file)
--- a/sdk/cwl/arvados_cwl/arvcontainer.py
+++ b/sdk/cwl/arvados_cwl/arvcontainer.py
@@ -37,6 +37,9 @@ from ._version import __version__
 logger = logging.getLogger('arvados.cwl-runner')
 metrics = logging.getLogger('arvados.cwl-runner.metrics')
 
+def cleanup_name_for_collection(name):
+    return name.replace("/", " ")
+
 class ArvadosContainer(JobBase):
     """Submit and manage a Crunch container request for executing a CWL CommandLineTool."""
 
@@ -262,7 +265,11 @@ class ArvadosContainer(JobBase):
         runtime_req, _ = self.get_requirement("http://arvados.org/cwl#RuntimeConstraints")
         if runtime_req:
             if "keep_cache" in runtime_req:
-                runtime_constraints["keep_cache_ram"] = math.ceil(runtime_req["keep_cache"] * 2**20)
+                if self.arvrunner.api.config()["Containers"].get("DefaultKeepCacheDisk", 0) > 0:
+                    # If DefaultKeepCacheDisk is non-zero it means we should use disk cache.
+                    runtime_constraints["keep_cache_disk"] = math.ceil(runtime_req["keep_cache"] * 2**20)
+                else:
+                    runtime_constraints["keep_cache_ram"] = math.ceil(runtime_req["keep_cache"] * 2**20)
             if "outputDirType" in runtime_req:
                 if runtime_req["outputDirType"] == "local_output_dir":
                     # Currently the default behavior.
@@ -320,7 +327,7 @@ class ArvadosContainer(JobBase):
         if runtimeContext.submit_runner_cluster:
             extra_submit_params["cluster_id"] = runtimeContext.submit_runner_cluster
 
-        container_request["output_name"] = "Output from step %s" % (self.name)
+        container_request["output_name"] = cleanup_name_for_collection("Output from step %s" % (self.name))
         container_request["output_ttl"] = self.output_ttl
         container_request["mounts"] = mounts
         container_request["secret_mounts"] = secret_mounts
@@ -450,7 +457,7 @@ class ArvadosContainer(JobBase):
 class RunnerContainer(Runner):
     """Submit and manage a container that runs arvados-cwl-runner."""
 
-    def arvados_job_spec(self, runtimeContext):
+    def arvados_job_spec(self, runtimeContext, git_info):
         """Create an Arvados container request for this workflow.
 
         The returned dict can be used to create a container passed as
@@ -512,7 +519,7 @@ class RunnerContainer(Runner):
                 "portable_data_hash": "%s" % workflowcollection
             }
         else:
-            packed = packed_workflow(self.arvrunner, self.embedded_tool, self.merged_map, runtimeContext)
+            packed = packed_workflow(self.arvrunner, self.embedded_tool, self.merged_map, runtimeContext, git_info)
             workflowpath = "/var/lib/cwl/workflow.json#main"
             container_req["mounts"]["/var/lib/cwl/workflow.json"] = {
                 "kind": "json",
@@ -521,6 +528,8 @@ class RunnerContainer(Runner):
             if self.embedded_tool.tool.get("id", "").startswith("arvwf:"):
                 container_req["properties"]["template_uuid"] = self.embedded_tool.tool["id"][6:33]
 
+        container_req["properties"].update({k.replace("http://arvados.org/cwl#", "arv:"): v for k, v in git_info.items()})
+
         properties_req, _ = self.embedded_tool.get_requirement("http://arvados.org/cwl#ProcessProperties")
         if properties_req:
             builder = make_builder(self.job_order, self.embedded_tool.hints, self.embedded_tool.requirements, runtimeContext, self.embedded_tool.metadata)
@@ -592,7 +601,7 @@ class RunnerContainer(Runner):
 
     def run(self, runtimeContext):
         runtimeContext.keepprefix = "keep:"
-        job_spec = self.arvados_job_spec(runtimeContext)
+        job_spec = self.arvados_job_spec(runtimeContext, self.git_info)
         if runtimeContext.project_uuid:
             job_spec["owner_uuid"] = runtimeContext.project_uuid