9766: Supports packing workflow, setting defaults, uploading dependent files/docker...
[arvados.git] / sdk / cwl / arvados_cwl / __init__.py
index 27af075f3652b4180e23906ed2fc914812036534..80734d2f7c8e41e45ce25b7bd38cba6094b071e3 100644 (file)
@@ -17,13 +17,13 @@ import cwltool.main
 import cwltool.workflow
 
 import arvados
 import cwltool.workflow
 
 import arvados
-import arvados.events
 import arvados.config
 
 from .arvcontainer import ArvadosContainer, RunnerContainer
 from .arvjob import ArvadosJob, RunnerJob, RunnerTemplate
 from .arvtool import ArvadosCommandTool
 from .fsaccess import CollectionFsAccess
 import arvados.config
 
 from .arvcontainer import ArvadosContainer, RunnerContainer
 from .arvjob import ArvadosJob, RunnerJob, RunnerTemplate
 from .arvtool import ArvadosCommandTool
 from .fsaccess import CollectionFsAccess
+from .arvworkflow import make_workflow
 
 from cwltool.process import shortname, UnsupportedRequirement
 from cwltool.pathmapper import adjustFileObjs
 
 from cwltool.process import shortname, UnsupportedRequirement
 from cwltool.pathmapper import adjustFileObjs
@@ -50,6 +50,8 @@ class ArvCwlRunner(object):
         self.num_retries = 4
         self.uuid = None
         self.work_api = work_api
         self.num_retries = 4
         self.uuid = None
         self.work_api = work_api
+        self.stop_polling = threading.Event()
+        self.poll_api = None
 
         if self.work_api is None:
             # todo: autodetect API to use.
 
         if self.work_api is None:
             # todo: autodetect API to use.
@@ -99,6 +101,41 @@ class ArvCwlRunner(object):
                     finally:
                         self.cond.release()
 
                     finally:
                         self.cond.release()
 
+    def poll_states(self):
+        """Poll status of jobs or containers listed in the processes dict.
+
+        Runs in a separate thread.
+        """
+
+        while True:
+            self.stop_polling.wait(15)
+            if self.stop_polling.is_set():
+                break
+            with self.lock:
+                keys = self.processes.keys()
+            if not keys:
+                continue
+
+            if self.work_api == "containers":
+                table = self.poll_api.containers()
+            elif self.work_api == "jobs":
+                table = self.poll_api.jobs()
+
+            try:
+                proc_states = table.list(filters=[["uuid", "in", keys]]).execute(num_retries=self.num_retries)
+            except Exception as e:
+                logger.warn("Error checking states on API server: %s", e)
+                continue
+
+            for p in proc_states["items"]:
+                self.on_message({
+                    "object_uuid": p["uuid"],
+                    "event_type": "update",
+                    "properties": {
+                        "new_attributes": p
+                    }
+                })
+
     def get_uploaded(self):
         return self.uploaded.copy()
 
     def get_uploaded(self):
         return self.uploaded.copy()
 
@@ -136,6 +173,9 @@ class ArvCwlRunner(object):
             # cwltool.main will write our return value to stdout.
             return tmpl.uuid
 
             # cwltool.main will write our return value to stdout.
             return tmpl.uuid
 
+        if kwargs.get("create_workflow") or kwargs.get("update_workflow"):
+            return make_workflow(self, tool, job_order, self.project_uuid, kwargs.get("update_workflow"))
+
         self.debug = kwargs.get("debug")
         self.ignore_docker_for_reuse = kwargs.get("ignore_docker_for_reuse")
 
         self.debug = kwargs.get("debug")
         self.ignore_docker_for_reuse = kwargs.get("ignore_docker_for_reuse")
 
@@ -182,12 +222,9 @@ class ArvCwlRunner(object):
             runnerjob.run()
             return runnerjob.uuid
 
             runnerjob.run()
             return runnerjob.uuid
 
-        arvados.config.settings()["ARVADOS_DISABLE_WEBSOCKETS"] = "1"
-
-        if self.work_api == "containers":
-            events = arvados.events.subscribe(arvados.api('v1'), [["object_uuid", "is_a", "arvados#container"]], self.on_message)
-        if self.work_api == "jobs":
-            events = arvados.events.subscribe(arvados.api('v1'), [["object_uuid", "is_a", "arvados#job"]], self.on_message)
+        self.poll_api = arvados.api('v1')
+        self.polling_thread = threading.Thread(target=self.poll_states)
+        self.polling_thread.start()
 
         if runnerjob:
             jobiter = iter((runnerjob,))
 
         if runnerjob:
             jobiter = iter((runnerjob,))
@@ -217,7 +254,6 @@ class ArvCwlRunner(object):
             while self.processes:
                 self.cond.wait(1)
 
             while self.processes:
                 self.cond.wait(1)
 
-            events.close()
         except UnsupportedRequirement:
             raise
         except:
         except UnsupportedRequirement:
             raise
         except:
@@ -233,6 +269,8 @@ class ArvCwlRunner(object):
                                                      body={"priority": "0"}).execute(num_retries=self.num_retries)
         finally:
             self.cond.release()
                                                      body={"priority": "0"}).execute(num_retries=self.num_retries)
         finally:
             self.cond.release()
+            self.stop_polling.set()
+            self.polling_thread.join()
 
         if self.final_status == "UnsupportedRequirement":
             raise UnsupportedRequirement("Check log for details.")
 
         if self.final_status == "UnsupportedRequirement":
             raise UnsupportedRequirement("Check log for details.")
@@ -301,6 +339,8 @@ def arg_parser():  # type: () -> argparse.ArgumentParser
     exgroup.add_argument("--local", action="store_false", help="Run workflow on local host (submits jobs to Arvados).",
                         default=True, dest="submit")
     exgroup.add_argument("--create-template", action="store_true", help="Create an Arvados pipeline template.")
     exgroup.add_argument("--local", action="store_false", help="Run workflow on local host (submits jobs to Arvados).",
                         default=True, dest="submit")
     exgroup.add_argument("--create-template", action="store_true", help="Create an Arvados pipeline template.")
+    exgroup.add_argument("--create-workflow", action="store_true", help="Create an Arvados workflow.")
+    exgroup.add_argument("--update-workflow", type=str, help="Update Arvados workflow.")
 
     exgroup = parser.add_mutually_exclusive_group()
     exgroup.add_argument("--wait", action="store_true", help="After submitting workflow runner job, wait for completion.",
 
     exgroup = parser.add_mutually_exclusive_group()
     exgroup.add_argument("--wait", action="store_true", help="After submitting workflow runner job, wait for completion.",
@@ -327,7 +367,7 @@ def main(args, stdout, stderr, api_client=None):
 
     job_order_object = None
     arvargs = parser.parse_args(args)
 
     job_order_object = None
     arvargs = parser.parse_args(args)
-    if arvargs.create_template and not arvargs.job_order:
+    if (arvargs.create_template or arvargs.create_workflow or arvargs.update_workflow) and not arvargs.job_order:
         job_order_object = ({}, "")
 
     try:
         job_order_object = ({}, "")
 
     try: