Merge branch 'master' into 10231-keep-cache-runtime-constraints

[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 5ad87d1ad0c29d2288816cf8952d24efbd1634e4..c2029b965b4f3b5173bcd981a67ff8bf3c809d3d 100644 (file)
--- a/sdk/cwl/arvados_cwl/arvcontainer.py
+++ b/sdk/cwl/arvados_cwl/arvcontainer.py
@@ -5,12 +5,13 @@ import os
 from cwltool.errors import WorkflowException
 from cwltool.process import get_feature, UnsupportedRequirement, shortname
 from cwltool.pathmapper import adjustFiles
+from cwltool.utils import aslist
 
 import arvados.collection
 
 from .arvdocker import arv_docker_get_image
 from . import done
-from .runner import Runner
+from .runner import Runner, arvados_jobs_image
 
 logger = logging.getLogger('arvados.cwl-runner')
 
@@ -42,30 +43,43 @@ class ArvadosContainer(object):
             }
         }
 
+        dirs = set()
         for f in self.pathmapper.files():
-            _, p = self.pathmapper.mapper(f)
-            mounts[p] = {
-                "kind": "collection",
-                "portable_data_hash": p[6:]
-            }
+            _, p, tp = self.pathmapper.mapper(f)
+            if tp == "Directory" and '/' not in p[6:]:
+                mounts[p] = {
+                    "kind": "collection",
+                    "portable_data_hash": p[6:]
+                }
+                dirs.add(p[6:])
+        for f in self.pathmapper.files():
+            _, p, tp = self.pathmapper.mapper(f)
+            if p[6:].split("/")[0] not in dirs:
+                mounts[p] = {
+                    "kind": "collection",
+                    "portable_data_hash": p[6:]
+                }
 
-        if self.generatefiles:
+        if self.generatefiles["listing"]:
             raise UnsupportedRequirement("Generate files not supported")
 
-        container_request["environment"] = {"TMPDIR": "/tmp"}
+        container_request["environment"] = {"TMPDIR": self.tmpdir, "HOME": self.outdir}
         if self.environment:
             container_request["environment"].update(self.environment)
 
         if self.stdin:
             raise UnsupportedRequirement("Stdin redirection currently not suppported")
 
+        if self.stderr:
+            raise UnsupportedRequirement("Stderr redirection currently not suppported")
+
         if self.stdout:
             mounts["stdout"] = {"kind": "file",
                                 "path": "%s/%s" % (self.outdir, self.stdout)}
 
         (docker_req, docker_is_req) = get_feature(self, "DockerRequirement")
         if not docker_req:
-            docker_req = {"dockerImageId": "arvados/jobs"}
+            docker_req = {"dockerImageId": arvados_jobs_image(self.arvrunner)}
 
         container_request["container_image"] = arv_docker_get_image(self.arvrunner.api,
                                                                      docker_req,
@@ -77,6 +91,18 @@ class ArvadosContainer(object):
             runtime_constraints["vcpus"] = resources.get("cores", 1)
             runtime_constraints["ram"] = resources.get("ram") * 2**20
 
+        api_req, _ = get_feature(self, "http://arvados.org/cwl#APIRequirement")
+        if api_req:
+            runtime_constraints["API"] = True
+
+        runtime_req, _ = get_feature(self, "http://arvados.org/cwl#RuntimeConstraints")
+        if runtime_req:
+            runtime_constraints["keep_cache_ram"] = runtime_req["keep_cache"]
+
+        partition_req, _ = get_feature(self, "http://arvados.org/cwl#PartitionRequirement")
+        if partition_req:
+            runtime_constraints["partition"] = aslist(partition_req["partition"])
+
         container_request["mounts"] = mounts
         container_request["runtime_constraints"] = runtime_constraints
 
@@ -87,10 +113,14 @@ class ArvadosContainer(object):
 
             self.arvrunner.processes[response["container_uuid"]] = self
 
-            logger.info("Container %s (%s) request state is %s", self.name, response["container_uuid"], response["state"])
+            container = self.arvrunner.api.containers().get(
+                uuid=response["container_uuid"]
+            ).execute(num_retries=self.arvrunner.num_retries)
+
+            logger.info("Container request %s (%s) state is %s with container %s %s", self.name, response["uuid"], response["state"], container["uuid"], container["state"])
 
-            if response["state"] == "Final":
-                self.done(response)
+            if container["state"] in ("Complete", "Cancelled"):
+                self.done(container)
         except Exception as e:
             logger.error("Got error %s" % str(e))
             self.output_callback({}, "permanentFail")
@@ -140,7 +170,9 @@ class RunnerContainer(Runner):
 
         workflowmapper = super(RunnerContainer, self).arvados_job_spec(dry_run=dry_run, pull_image=pull_image, **kwargs)
 
-        with arvados.collection.Collection(api_client=self.arvrunner.api) as jobobj:
+        with arvados.collection.Collection(api_client=self.arvrunner.api,
+                                           keep_client=self.arvrunner.keep_client,
+                                           num_retries=self.arvrunner.num_retries) as jobobj:
             with jobobj.open("cwl.input.json", "w") as f:
                 json.dump(self.job_order, f, sort_keys=True, indent=4)
             jobobj.save_new(owner_uuid=self.arvrunner.project_uuid)
@@ -151,20 +183,20 @@ class RunnerContainer(Runner):
         workflowcollection = workflowcollection[5:workflowcollection.index('/')]
         jobpath = "/var/lib/cwl/job/cwl.input.json"
 
-        container_image = arv_docker_get_image(self.arvrunner.api,
-                                               {"dockerImageId": "arvados/jobs"},
-                                               pull_image,
-                                               self.arvrunner.project_uuid)
+        command = ["arvados-cwl-runner", "--local", "--api=containers"]
+        if self.output_name:
+            command.append("--output-name=" + self.output_name)
+        command.extend([workflowpath, jobpath])
 
         return {
-            "command": ["arvados-cwl-runner", "--local", "--api=containers", workflowpath, jobpath],
+            "command": command,
             "owner_uuid": self.arvrunner.project_uuid,
             "name": self.name,
             "output_path": "/var/spool/cwl",
             "cwd": "/var/spool/cwl",
             "priority": 1,
             "state": "Committed",
-            "container_image": container_image,
+            "container_image": arvados_jobs_image(self.arvrunner),
             "mounts": {
                 "/var/lib/cwl/workflow": {
                     "kind": "collection",