metaflow 2.12.21__tar.gz → 2.12.22__tar.gz

{metaflow-2.12.21/metaflow.egg-info → metaflow-2.12.22}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: metaflow
-Version: 2.12.21
+Version: 2.12.22
 Summary: Metaflow: More Data Science, Less Engineering
 Author: Metaflow Developers
 Author-email: help@metaflow.org
@@ -26,7 +26,7 @@ License-File: LICENSE
 Requires-Dist: requests
 Requires-Dist: boto3
 Provides-Extra: stubs
-Requires-Dist: metaflow-stubs==2.12.21; extra == "stubs"
+Requires-Dist: metaflow-stubs==2.12.22; extra == "stubs"
 
 ![Metaflow_Logo_Horizontal_FullColor_Ribbon_Dark_RGB](https://user-images.githubusercontent.com/763451/89453116-96a57e00-d713-11ea-9fa6-82b29d4d6eff.png)
 

{metaflow-2.12.21 → metaflow-2.12.22}/metaflow/plugins/argo/argo_workflows.py

@@ -455,11 +455,17 @@ class ArgoWorkflows(object):
                 )
             seen.add(norm)
 
-            if param.kwargs.get("type") == JSONType or isinstance(
-                param.kwargs.get("type"), FilePathClass
-            ):
-                # Special-case this to avoid touching core
+            extra_attrs = {}
+            if param.kwargs.get("type") == JSONType:
+                param_type = str(param.kwargs.get("type").name)
+            elif isinstance(param.kwargs.get("type"), FilePathClass):
                 param_type = str(param.kwargs.get("type").name)
+                extra_attrs["is_text"] = getattr(
+                    param.kwargs.get("type"), "_is_text", True
+                )
+                extra_attrs["encoding"] = getattr(
+                    param.kwargs.get("type"), "_encoding", "utf-8"
+                )
             else:
                 param_type = str(param.kwargs.get("type").__name__)
 
@@ -487,6 +493,7 @@ class ArgoWorkflows(object):
                 type=param_type,
                 description=param.kwargs.get("help"),
                 is_required=is_required,
+                **extra_attrs
             )
         return parameters
 
@@ -2515,10 +2522,29 @@ class ArgoWorkflows(object):
         # Use all the affordances available to _parameters task
         executable = self.environment.executable("_parameters")
         run_id = "argo-{{workflow.name}}"
-        entrypoint = [executable, "-m metaflow.plugins.argo.daemon"]
-        heartbeat_cmds = "{entrypoint} --flow_name {flow_name} --run_id {run_id} {tags} heartbeat".format(
+        script_name = os.path.basename(sys.argv[0])
+        entrypoint = [executable, script_name]
+        # FlowDecorators can define their own top-level options. These might affect run level information
+        # so it is important to pass these to the heartbeat process as well, as it might be the first task to register a run.
+        top_opts_dict = {}
+        for deco in flow_decorators(self.flow):
+            top_opts_dict.update(deco.get_top_level_options())
+
+        top_level = list(dict_to_cli_options(top_opts_dict)) + [
+            "--quiet",
+            "--metadata=%s" % self.metadata.TYPE,
+            "--environment=%s" % self.environment.TYPE,
+            "--datastore=%s" % self.flow_datastore.TYPE,
+            "--datastore-root=%s" % self.flow_datastore.datastore_root,
+            "--event-logger=%s" % self.event_logger.TYPE,
+            "--monitor=%s" % self.monitor.TYPE,
+            "--no-pylint",
+            "--with=argo_workflows_internal:auto-emit-argo-events=%i"
+            % self.auto_emit_argo_events,
+        ]
+        heartbeat_cmds = "{entrypoint} {top_level} argo-workflows heartbeat --run_id {run_id} {tags}".format(
             entrypoint=" ".join(entrypoint),
-            flow_name=self.flow.name,
+            top_level=" ".join(top_level) if top_level else "",
             run_id=run_id,
             tags=" ".join(["--tag %s" % t for t in self.tags]) if self.tags else "",
         )
@@ -2569,12 +2595,16 @@ class ArgoWorkflows(object):
             "METAFLOW_SERVICE_URL": SERVICE_INTERNAL_URL,
             "METAFLOW_SERVICE_HEADERS": json.dumps(SERVICE_HEADERS),
             "METAFLOW_USER": "argo-workflows",
+            "METAFLOW_DATASTORE_SYSROOT_S3": DATASTORE_SYSROOT_S3,
+            "METAFLOW_DATATOOLS_S3ROOT": DATATOOLS_S3ROOT,
             "METAFLOW_DEFAULT_DATASTORE": self.flow_datastore.TYPE,
             "METAFLOW_DEFAULT_METADATA": DEFAULT_METADATA,
+            "METAFLOW_CARD_S3ROOT": CARD_S3ROOT,
             "METAFLOW_KUBERNETES_WORKLOAD": 1,
+            "METAFLOW_KUBERNETES_FETCH_EC2_METADATA": KUBERNETES_FETCH_EC2_METADATA,
             "METAFLOW_RUNTIME_ENVIRONMENT": "kubernetes",
             "METAFLOW_OWNER": self.username,
-            "METAFLOW_PRODUCTION_TOKEN": self.production_token,
+            "METAFLOW_PRODUCTION_TOKEN": self.production_token,  # Used in identity resolving. This affects system tags.
         }
         # support Metaflow sandboxes
         env["METAFLOW_INIT_SCRIPT"] = KUBERNETES_SANDBOX_INIT_SCRIPT

{metaflow-2.12.21 → metaflow-2.12.22}/metaflow/plugins/argo/argo_workflows_cli.py

@@ -4,6 +4,7 @@ import platform
 import re
 import sys
 from hashlib import sha1
+from time import sleep
 
 from metaflow import JSONType, Run, current, decorators, parameters
 from metaflow._vendor import click
@@ -959,6 +960,31 @@ def list_workflow_templates(obj, all=None):
         obj.echo_always(template_name)
 
 
+# Internal CLI command to run a heartbeat daemon in an Argo Workflows Daemon container.
+@argo_workflows.command(hidden=True, help="start heartbeat process for a run")
+@click.option("--run_id", required=True)
+@click.option(
+    "--tag",
+    "tags",
+    multiple=True,
+    default=None,
+    help="Annotate all objects produced by Argo Workflows runs "
+    "with the given tag. You can specify this option multiple "
+    "times to attach multiple tags.",
+)
+@click.pass_obj
+def heartbeat(obj, run_id, tags=None):
+    # Try to register a run in case the start task has not taken care of it yet.
+    obj.metadata.register_run_id(run_id, tags)
+    # Start run heartbeat
+    obj.metadata.start_run_heartbeat(obj.flow.name, run_id)
+    # Keepalive loop
+    while True:
+        # Do not pollute daemon logs with anything unnecessary,
+        # as they might be extremely long running.
+        sleep(10)
+
+
 def validate_run_id(
     workflow_name, token_prefix, authorize, run_id, instructions_fn=None
 ):

metaflow-2.12.22/metaflow/version.py

@@ -0,0 +1 @@
+metaflow_version = "2.12.22"

{metaflow-2.12.21 → metaflow-2.12.22/metaflow.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: metaflow
-Version: 2.12.21
+Version: 2.12.22
 Summary: Metaflow: More Data Science, Less Engineering
 Author: Metaflow Developers
 Author-email: help@metaflow.org
@@ -26,7 +26,7 @@ License-File: LICENSE
 Requires-Dist: requests
 Requires-Dist: boto3
 Provides-Extra: stubs
-Requires-Dist: metaflow-stubs==2.12.21; extra == "stubs"
+Requires-Dist: metaflow-stubs==2.12.22; extra == "stubs"
 
 ![Metaflow_Logo_Horizontal_FullColor_Ribbon_Dark_RGB](https://user-images.githubusercontent.com/763451/89453116-96a57e00-d713-11ea-9fa6-82b29d4d6eff.png)
 

{metaflow-2.12.21 → metaflow-2.12.22}/metaflow.egg-info/SOURCES.txt

@@ -191,7 +191,6 @@ metaflow/plugins/argo/argo_workflows_cli.py
 metaflow/plugins/argo/argo_workflows_decorator.py
 metaflow/plugins/argo/argo_workflows_deployer.py
 metaflow/plugins/argo/capture_error.py
-metaflow/plugins/argo/daemon.py
 metaflow/plugins/argo/generate_input_paths.py
 metaflow/plugins/argo/jobset_input_paths.py
 metaflow/plugins/aws/__init__.py

{metaflow-2.12.21 → metaflow-2.12.22}/metaflow.egg-info/requires.txt

@@ -2,4 +2,4 @@ requests
 boto3
 
 [stubs]
-metaflow-stubs==2.12.21
+metaflow-stubs==2.12.22

metaflow-2.12.21/metaflow/plugins/argo/daemon.py

@@ -1,59 +0,0 @@
-from collections import namedtuple
-from time import sleep
-from metaflow.metaflow_config import DEFAULT_METADATA
-from metaflow.metaflow_environment import MetaflowEnvironment
-from metaflow.plugins import METADATA_PROVIDERS
-from metaflow._vendor import click
-
-
-class CliState:
-    pass
-
-
-@click.group()
-@click.option("--flow_name", required=True)
-@click.option("--run_id", required=True)
-@click.option(
-    "--tag",
-    "tags",
-    multiple=True,
-    default=None,
-    help="Annotate all objects produced by Argo Workflows runs "
-    "with the given tag. You can specify this option multiple "
-    "times to attach multiple tags.",
-)
-@click.pass_context
-def cli(ctx, flow_name, run_id, tags=None):
-    ctx.obj = CliState()
-    ctx.obj.flow_name = flow_name
-    ctx.obj.run_id = run_id
-    ctx.obj.tags = tags
-    # Use a dummy flow to initialize the environment and metadata service,
-    # as we only need a name for the flow object.
-    flow = namedtuple("DummyFlow", "name")
-    dummyflow = flow(flow_name)
-
-    # Initialize a proper metadata service instance
-    environment = MetaflowEnvironment(dummyflow)
-
-    ctx.obj.metadata = [m for m in METADATA_PROVIDERS if m.TYPE == DEFAULT_METADATA][0](
-        environment, dummyflow, None, None
-    )
-
-
-@cli.command(help="start heartbeat process for a run")
-@click.pass_obj
-def heartbeat(obj):
-    # Try to register a run in case the start task has not taken care of it yet.
-    obj.metadata.register_run_id(obj.run_id, obj.tags)
-    # Start run heartbeat
-    obj.metadata.start_run_heartbeat(obj.flow_name, obj.run_id)
-    # Keepalive loop
-    while True:
-        # Do not pollute daemon logs with anything unnecessary,
-        # as they might be extremely long running.
-        sleep(10)
-
-
-if __name__ == "__main__":
-    cli()

metaflow-2.12.21/metaflow/version.py

@@ -1 +0,0 @@
-metaflow_version = "2.12.21"