metaflow 2.11.14__tar.gz → 2.11.15__tar.gz

{metaflow-2.11.14/metaflow.egg-info → metaflow-2.11.15}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: metaflow
-Version: 2.11.14
+Version: 2.11.15
 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==2.11.14; extra == "stubs"
+Requires-Dist: metaflow-stubs==2.11.15; extra == "stubs"
 
 ![Metaflow_Logo_Horizontal_FullColor_Ribbon_Dark_RGB](https://user-images.githubusercontent.com/763451/89453116-96a57e00-d713-11ea-9fa6-82b29d4d6eff.png)
 

{metaflow-2.11.14 → metaflow-2.11.15}/metaflow/cli.py

@@ -287,126 +287,6 @@ def dump(obj, input_path, private=None, max_value_size=None, include=None, file=
         echo("Artifacts written to *%s*" % file)
 
 
-@cli.command(
-    help="Show stdout/stderr produced by a task or all tasks in a step. "
-    "The format for input-path is either <run_id>/<step_name> or "
-    "<run_id>/<step_name>/<task_id>."
-)
-@click.argument("input-path")
-@click.option(
-    "--stdout/--no-stdout",
-    default=False,
-    show_default=True,
-    help="Show stdout of the task.",
-)
-@click.option(
-    "--stderr/--no-stderr",
-    default=False,
-    show_default=True,
-    help="Show stderr of the task.",
-)
-@click.option(
-    "--both/--no-both",
-    default=True,
-    show_default=True,
-    help="Show both stdout and stderr of the task.",
-)
-@click.option(
-    "--timestamps/--no-timestamps",
-    default=False,
-    show_default=True,
-    help="Show timestamps.",
-)
-@click.pass_obj
-def logs(obj, input_path, stdout=None, stderr=None, both=None, timestamps=False):
-    types = set()
-    if stdout:
-        types.add("stdout")
-        both = False
-    if stderr:
-        types.add("stderr")
-        both = False
-    if both:
-        types.update(("stdout", "stderr"))
-
-    streams = list(sorted(types, reverse=True))
-
-    # Pathspec can either be run_id/step_name or run_id/step_name/task_id.
-    parts = input_path.split("/")
-    if len(parts) == 2:
-        run_id, step_name = parts
-        task_id = None
-    elif len(parts) == 3:
-        run_id, step_name, task_id = parts
-    else:
-        raise CommandException(
-            "input_path should either be run_id/step_name "
-            "or run_id/step_name/task_id"
-        )
-
-    datastore_set = TaskDataStoreSet(
-        obj.flow_datastore, run_id, steps=[step_name], allow_not_done=True
-    )
-    if task_id:
-        ds_list = [
-            TaskDataStore(
-                obj.flow_datastore,
-                run_id=run_id,
-                step_name=step_name,
-                task_id=task_id,
-                mode="r",
-                allow_not_done=True,
-            )
-        ]
-    else:
-        ds_list = list(datastore_set)  # get all tasks
-
-    if ds_list:
-
-        def echo_unicode(line, **kwargs):
-            click.secho(line.decode("UTF-8", errors="replace"), **kwargs)
-
-        # old style logs are non mflog-style logs
-        maybe_old_style = True
-        for ds in ds_list:
-            echo(
-                "Dumping logs of run_id=*{run_id}* "
-                "step=*{step}* task_id=*{task_id}*".format(
-                    run_id=ds.run_id, step=ds.step_name, task_id=ds.task_id
-                ),
-                fg="magenta",
-            )
-
-            for stream in streams:
-                echo(stream, bold=True)
-                logs = ds.load_logs(LOG_SOURCES, stream)
-                if any(data for _, data in logs):
-                    # attempt to read new, mflog-style logs
-                    for line in mflog.merge_logs([blob for _, blob in logs]):
-                        if timestamps:
-                            ts = mflog.utc_to_local(line.utc_tstamp)
-                            tstamp = ts.strftime("%Y-%m-%d %H:%M:%S.%f")[:-3]
-                            click.secho(tstamp + " ", fg=LOGGER_TIMESTAMP, nl=False)
-                        echo_unicode(line.msg)
-                    maybe_old_style = False
-                elif maybe_old_style:
-                    # if they are not available, we may be looking at
-                    # a legacy run (unless we have seen new-style data already
-                    # for another stream). This return an empty string if
-                    # nothing is found
-                    log = ds.load_log_legacy(stream)
-                    if log and timestamps:
-                        raise CommandException(
-                            "We can't show --timestamps for old runs. Sorry!"
-                        )
-                    echo_unicode(log, nl=False)
-    else:
-        raise CommandException(
-            "No Tasks found at the given path -- "
-            "either none exist or none have started yet"
-        )
-
-
 # TODO - move step and init under a separate 'internal' subcommand
 
 

{metaflow-2.11.14 → metaflow-2.11.15}/metaflow/datastore/datastore_set.py

@@ -22,7 +22,7 @@ class TaskDataStoreSet(object):
         prefetch_data_artifacts=None,
         allow_not_done=False,
     ):
-        self.task_datastores = flow_datastore.get_latest_task_datastores(
+        self.task_datastores = flow_datastore.get_task_datastores(
             run_id, steps=steps, pathspecs=pathspecs, allow_not_done=allow_not_done
         )
 

{metaflow-2.11.14 → metaflow-2.11.15}/metaflow/datastore/flow_datastore.py

@@ -67,8 +67,15 @@ class FlowDataStore(object):
     def datastore_root(self):
         return self._storage_impl.datastore_root
 
-    def get_latest_task_datastores(
-        self, run_id=None, steps=None, pathspecs=None, allow_not_done=False
+    def get_task_datastores(
+        self,
+        run_id=None,
+        steps=None,
+        pathspecs=None,
+        allow_not_done=False,
+        attempt=None,
+        include_prior=False,
+        mode="r",
     ):
         """
         Return a list of TaskDataStore for a subset of the tasks.
@@ -93,6 +100,12 @@ class FlowDataStore(object):
         allow_not_done : bool, optional
             If True, returns the latest attempt of a task even if that attempt
             wasn't marked as done, by default False
+        attempt : int, optional
+            Attempt number of the tasks to return.  If not provided, returns latest attempt.
+        include_prior : boolean, default False
+            If True, returns all attempts up to and including attempt.
+        mode : str, default "r"
+            Mode to initialize the returned TaskDataStores in.
 
         Returns
         -------
@@ -126,8 +139,13 @@ class FlowDataStore(object):
                 if task.is_file is False
             ]
         urls = []
+        # parse content urls for specific attempt only, or for all attempts in max range
+        attempt_range = range(metaflow_config.MAX_ATTEMPTS)
+        # we have no reason to check for attempts greater than MAX_ATTEMPTS, as they do not exist.
+        if attempt is not None and attempt <= metaflow_config.MAX_ATTEMPTS - 1:
+            attempt_range = range(attempt + 1) if include_prior else [attempt]
         for task_url in task_urls:
-            for attempt in range(metaflow_config.MAX_ATTEMPTS):
+            for attempt in attempt_range:
                 for suffix in [
                     TaskDataStore.METADATA_DATA_SUFFIX,
                     TaskDataStore.METADATA_ATTEMPT_SUFFIX,
@@ -168,11 +186,19 @@ class FlowDataStore(object):
             for (run, step, task), attempt in latest_started_attempts.items()
         )
         if allow_not_done:
-            latest_to_fetch = latest_started_attempts
+            latest_to_fetch = (
+                done_attempts.union(latest_started_attempts)
+                if include_prior
+                else latest_started_attempts
+            )
         else:
-            latest_to_fetch = latest_started_attempts & done_attempts
+            latest_to_fetch = (
+                done_attempts
+                if include_prior
+                else (latest_started_attempts & done_attempts)
+            )
         latest_to_fetch = [
-            (v[0], v[1], v[2], v[3], data_objs.get(v), "r", allow_not_done)
+            (v[0], v[1], v[2], v[3], data_objs.get(v), mode, allow_not_done)
             for v in latest_to_fetch
         ]
         return list(itertools.starmap(self.get_task_datastore, latest_to_fetch))

{metaflow-2.11.14 → metaflow-2.11.15}/metaflow/datastore/task_datastore.py

@@ -173,6 +173,26 @@ class TaskDataStore(object):
                 if data_obj is not None:
                     self._objects = data_obj.get("objects", {})
                     self._info = data_obj.get("info", {})
+        elif self._mode == "d":
+            self._objects = {}
+            self._info = {}
+
+            if self._attempt is None:
+                for i in range(metaflow_config.MAX_ATTEMPTS):
+                    check_meta = self._metadata_name_for_attempt(
+                        self.METADATA_ATTEMPT_SUFFIX, i
+                    )
+                    if self.has_metadata(check_meta, add_attempt=False):
+                        self._attempt = i
+
+            # Do not allow destructive operations on the datastore if attempt is still in flight
+            # and we explicitly did not allow operating on running tasks.
+            if not allow_not_done and not self.has_metadata(self.METADATA_DONE_SUFFIX):
+                raise DataException(
+                    "No completed attempts of the task was found for task '%s'"
+                    % self._path
+                )
+
         else:
             raise DataException("Unknown datastore mode: '%s'" % self._mode)
 
@@ -750,6 +770,36 @@ class TaskDataStore(object):
                 to_store_dict[n] = data
         self._save_file(to_store_dict)
 
+    @require_mode("d")
+    def scrub_logs(self, logsources, stream, attempt_override=None):
+        path_logsources = {
+            self._metadata_name_for_attempt(
+                self._get_log_location(s, stream),
+                attempt_override=attempt_override,
+            ): s
+            for s in logsources
+        }
+
+        # Legacy log paths
+        legacy_log = self._metadata_name_for_attempt(
+            "%s.log" % stream, attempt_override
+        )
+        path_logsources[legacy_log] = stream
+
+        existing_paths = [
+            path
+            for path in path_logsources.keys()
+            if self.has_metadata(path, add_attempt=False)
+        ]
+
+        # Replace log contents with [REDACTED source stream]
+        to_store_dict = {
+            path: bytes("[REDACTED %s %s]" % (path_logsources[path], stream), "utf-8")
+            for path in existing_paths
+        }
+
+        self._save_file(to_store_dict, add_attempt=False, allow_overwrite=True)
+
     @require_mode("r")
     def load_log_legacy(self, stream, attempt_override=None):
         """

{metaflow-2.11.14 → metaflow-2.11.15}/metaflow/plugins/__init__.py

@@ -14,6 +14,7 @@ CLIS_DESC = [
     ("argo-workflows", ".argo.argo_workflows_cli.cli"),
     ("card", ".cards.card_cli.cli"),
     ("tag", ".tag_cli.cli"),
+    ("logs", ".logs_cli.cli"),
 ]
 
 from .test_unbounded_foreach_decorator import InternalTestUnboundedForeachInput

{metaflow-2.11.14 → metaflow-2.11.15}/metaflow/plugins/aws/batch/batch_cli.py

@@ -10,7 +10,7 @@ from metaflow.exception import CommandException, METAFLOW_EXIT_DISALLOW_RETRY
 from metaflow.metadata.util import sync_local_metadata_from_datastore
 from metaflow.metaflow_config import DATASTORE_LOCAL_DIR
 from metaflow.mflog import TASK_LOG_SOURCE
-
+from metaflow.unbounded_foreach import UBF_CONTROL, UBF_TASK
 from .batch import Batch, BatchKilledException
 
 
@@ -150,8 +150,10 @@ def kill(ctx, run_id, user, my_runs):
 @click.option("--tmpfs-tempdir", is_flag=True, help="tmpfs requirement for AWS Batch.")
 @click.option("--tmpfs-size", help="tmpfs requirement for AWS Batch.")
 @click.option("--tmpfs-path", help="tmpfs requirement for AWS Batch.")
-# TODO: Maybe remove it altogether since it's not used here
-@click.option("--ubf-context", default=None, type=click.Choice([None, "ubf_control"]))
+# NOTE: ubf-context is not explicitly used, but @parallel decorator tries to pass this so keep it for now
+@click.option(
+    "--ubf-context", default=None, type=click.Choice(["none", UBF_CONTROL, UBF_TASK])
+)
 @click.option("--host-volumes", multiple=True)
 @click.option("--efs-volumes", multiple=True)
 @click.option(
@@ -344,7 +346,7 @@ def step(
                 log_options=log_options,
                 num_parallel=num_parallel,
             )
-    except Exception as e:
+    except Exception:
         traceback.print_exc()
         _sync_metadata()
         sys.exit(METAFLOW_EXIT_DISALLOW_RETRY)

{metaflow-2.11.14 → metaflow-2.11.15}/metaflow/plugins/datatools/s3/s3.py

@@ -1245,12 +1245,12 @@ class S3(object):
 
         def _store():
             for key_obj in key_objs:
-                if isinstance(key_obj, tuple):
-                    key = key_obj[0]
-                    obj = key_obj[1]
-                else:
+                if isinstance(key_obj, S3PutObject):
                     key = key_obj.key
                     obj = key_obj.value
+                else:
+                    key = key_obj[0]
+                    obj = key_obj[1]
                 store_info = {
                     "key": key,
                     "content_type": getattr(key_obj, "content_type", None),
@@ -1319,12 +1319,12 @@ class S3(object):
 
         def _check():
             for key_path in key_paths:
-                if isinstance(key_path, tuple):
-                    key = key_path[0]
-                    path = key_path[1]
-                else:
+                if isinstance(key_path, S3PutObject):
                     key = key_path.key
                     path = key_path.path
+                else:
+                    key = key_path[0]
+                    path = key_path[1]
                 store_info = {
                     "key": key,
                     "content_type": getattr(key_path, "content_type", None),