mlrun 1.7.0rc14__py3-none-any.whl → 1.7.0rc16__py3-none-any.whl

mlrun/utils/helpers.py

@@ -39,7 +39,7 @@ import pandas
 import semver
 import yaml
 from dateutil import parser
-from deprecated import deprecated
+from mlrun_pipelines.models import PipelineRun
 from pandas._libs.tslibs.timestamps import Timedelta, Timestamp
 from yaml.representer import RepresenterError
 
@@ -76,19 +76,6 @@ class OverwriteBuildParamsWarning(FutureWarning):
     pass
 
 
-# TODO: remove in 1.7.0
-@deprecated(
-    version="1.5.0",
-    reason="'parse_versioned_object_uri' will be removed from this file in 1.7.0, use "
-    "'mlrun.common.helpers.parse_versioned_object_uri' instead",
-    category=FutureWarning,
-)
-def parse_versioned_object_uri(uri: str, default_project: str = ""):
-    return mlrun.common.helpers.parse_versioned_object_uri(
-        uri=uri, default_project=default_project
-    )
-
-
 class StorePrefix:
     """map mlrun store objects to prefixes"""
 
@@ -119,14 +106,9 @@ class StorePrefix:
 
 
 def get_artifact_target(item: dict, project=None):
-    if is_legacy_artifact(item):
-        db_key = item.get("db_key")
-        project_str = project or item.get("project")
-        tree = item.get("tree")
-    else:
-        db_key = item["spec"].get("db_key")
-        project_str = project or item["metadata"].get("project")
-        tree = item["metadata"].get("tree")
+    db_key = item["spec"].get("db_key")
+    project_str = project or item["metadata"].get("project")
+    tree = item["metadata"].get("tree")
 
     kind = item.get("kind")
     if kind in ["dataset", "model", "artifact"] and db_key:
@@ -135,11 +117,15 @@ def get_artifact_target(item: dict, project=None):
             target = f"{target}@{tree}"
         return target
 
-    return (
-        item.get("target_path")
-        if is_legacy_artifact(item)
-        else item["spec"].get("target_path")
-    )
+    return item["spec"].get("target_path")
+
+
+# TODO: left for migrations testing purposes. Remove in 1.8.0.
+def is_legacy_artifact(artifact):
+    if isinstance(artifact, dict):
+        return "metadata" not in artifact
+    else:
+        return not hasattr(artifact, "metadata")
 
 
 logger = create_logger(config.log_level, config.log_formatter, "mlrun", sys.stdout)
@@ -801,34 +787,6 @@ def gen_html_table(header, rows=None):
     return style + '<table class="tg">\n' + out + "</table>\n\n"
 
 
-def new_pipe_metadata(
-    artifact_path: str = None,
-    cleanup_ttl: int = None,
-    op_transformers: list[typing.Callable] = None,
-):
-    from kfp.dsl import PipelineConf
-
-    def _set_artifact_path(task):
-        from kubernetes import client as k8s_client
-
-        task.add_env_variable(
-            k8s_client.V1EnvVar(name="MLRUN_ARTIFACT_PATH", value=artifact_path)
-        )
-        return task
-
-    conf = PipelineConf()
-    cleanup_ttl = cleanup_ttl or int(config.kfp_ttl)
-
-    if cleanup_ttl:
-        conf.set_ttl_seconds_after_finished(cleanup_ttl)
-    if artifact_path:
-        conf.add_op_transformer(_set_artifact_path)
-    if op_transformers:
-        for op_transformer in op_transformers:
-            conf.add_op_transformer(op_transformer)
-    return conf
-
-
 def _convert_python_package_version_to_image_tag(version: typing.Optional[str]):
     return (
         version.replace("+", "-").replace("0.0.0-", "") if version is not None else None
@@ -1018,14 +976,15 @@ def get_ui_url(project, uid=None):
 def get_workflow_url(project, id=None):
     url = ""
     if mlrun.mlconf.resolve_ui_url():
-        url = "{}/{}/{}/jobs/monitor-workflows/workflow/{}".format(
-            mlrun.mlconf.resolve_ui_url(), mlrun.mlconf.ui.projects_prefix, project, id
+        url = (
+            f"{mlrun.mlconf.resolve_ui_url()}/{mlrun.mlconf.ui.projects_prefix}"
+            f"/{project}/jobs/monitor-workflows/workflow/{id}"
         )
     return url
 
 
 def are_strings_in_exception_chain_messages(
-    exception: Exception, strings_list=list[str]
+    exception: Exception, strings_list: list[str]
 ) -> bool:
     while exception is not None:
         if any([string in str(exception) for string in strings_list]):
@@ -1291,13 +1250,6 @@ def str_to_timestamp(time_str: str, now_time: Timestamp = None):
     return Timestamp(time_str)
 
 
-def is_legacy_artifact(artifact):
-    if isinstance(artifact, dict):
-        return "metadata" not in artifact
-    else:
-        return not hasattr(artifact, "metadata")
-
-
 def is_link_artifact(artifact):
     if isinstance(artifact, dict):
         return (
@@ -1307,7 +1259,7 @@ def is_link_artifact(artifact):
         return artifact.kind == mlrun.common.schemas.ArtifactCategories.link.value
 
 
-def format_run(run: dict, with_project=False) -> dict:
+def format_run(run: PipelineRun, with_project=False) -> dict:
     fields = [
         "id",
         "name",
@@ -1344,17 +1296,17 @@ def format_run(run: dict, with_project=False) -> dict:
     # pipelines are yet to populate the status or workflow has failed
     # as observed https://jira.iguazeng.com/browse/ML-5195
     # set to unknown to ensure a status is returned
-    if run["status"] is None:
-        run["status"] = inflection.titleize(mlrun.runtimes.constants.RunStates.unknown)
+    if run.get("status", None) is None:
+        run["status"] = inflection.titleize(
+            mlrun.common.runtimes.constants.RunStates.unknown
+        )
 
     return run
 
 
 def get_in_artifact(artifact: dict, key, default=None, raise_on_missing=False):
     """artifact can be dict or Artifact object"""
-    if is_legacy_artifact(artifact):
-        return artifact.get(key, default)
-    elif key == "kind":
+    if key == "kind":
         return artifact.get(key, default)
     else:
         for block in ["metadata", "spec", "status"]:
@@ -1596,3 +1548,52 @@ def get_serving_spec():
             )
     spec = json.loads(data)
     return spec
+
+
+def additional_filters_warning(additional_filters, class_name):
+    if additional_filters and any(additional_filters):
+        mlrun.utils.logger.warn(
+            f"additional_filters parameter is not supported in {class_name},"
+            f" parameter has been ignored."
+        )
+
+
+def validate_component_version_compatibility(
+    component_name: typing.Literal["iguazio", "nuclio"], *min_versions: str
+):
+    """
+    :param component_name: Name of the component to validate compatibility for.
+    :param min_versions: Valid minimum version(s) required, assuming no 2 versions has equal major and minor.
+    """
+    parsed_min_versions = [
+        semver.VersionInfo.parse(min_version) for min_version in min_versions
+    ]
+    parsed_current_version = None
+    component_current_version = None
+    try:
+        if component_name == "iguazio":
+            parsed_current_version = mlrun.mlconf.get_parsed_igz_version()
+            component_current_version = mlrun.mlconf.igz_version
+        if component_name == "nuclio":
+            parsed_current_version = semver.VersionInfo.parse(
+                mlrun.mlconf.nuclio_version
+            )
+            component_current_version = mlrun.mlconf.nuclio_version
+        if not parsed_current_version:
+            return True
+    except ValueError:
+        # only log when version is set but invalid
+        if component_current_version:
+            logger.warning(
+                "Unable to parse current version, assuming compatibility",
+                component_name=component_name,
+                current_version=component_current_version,
+                min_versions=min_versions,
+            )
+        return True
+
+    parsed_min_versions.sort(reverse=True)
+    for parsed_min_version in parsed_min_versions:
+        if parsed_current_version < parsed_min_version:
+            return False
+    return True

mlrun/utils/notifications/notification/base.py

@@ -77,7 +77,7 @@ class NotificationBase:
                 return f"[{severity}] {message}"
             return (
                 f"[{severity}] {message} for project {alert.project} "
-                f"UID {event_data.entity.id}. Value {event_data.value}"
+                f"UID {event_data.entity.ids[0]}. Values {event_data.value_dict}"
             )
 
         if not runs:

mlrun/utils/notifications/notification/slack.py

@@ -135,7 +135,7 @@ class SlackNotification(NotificationBase):
         line = [
             self._get_slack_row(f":bell: {alert.name} alert has occurred"),
             self._get_slack_row(f"*Project:*\n{alert.project}"),
-            self._get_slack_row(f"*UID:*\n{event_data.entity.id}"),
+            self._get_slack_row(f"*UID:*\n{event_data.entity.ids[0]}"),
         ]
         if event_data.value_dict:
             data_lines = []
@@ -144,7 +144,9 @@ class SlackNotification(NotificationBase):
             data_text = "\n".join(data_lines)
             line.append(self._get_slack_row(f"*Event data:*\n{data_text}"))
 
-        if url := mlrun.utils.helpers.get_ui_url(alert.project, event_data.entity.id):
+        if url := mlrun.utils.helpers.get_ui_url(
+            alert.project, event_data.entity.ids[0]
+        ):
             line.append(self._get_slack_row(f"*Overview:*\n<{url}|*Job overview*>"))
 
         return line
@@ -152,24 +154,29 @@ class SlackNotification(NotificationBase):
     def _get_run_line(self, run: dict) -> dict:
         meta = run["metadata"]
         url = mlrun.utils.helpers.get_ui_url(meta.get("project"), meta.get("uid"))
-        if url:
+
+        # Only show the URL if the run is not a function (serving or mlrun function)
+        kind = run.get("step_kind")
+        if url and not kind or kind == "run":
             line = f'<{url}|*{meta.get("name")}*>'
         else:
             line = meta.get("name")
         state = run["status"].get("state", "")
+        if kind:
+            line = f'{line} *({run.get("step_kind", run.get("kind", ""))})*'
         line = f'{self.emojis.get(state, ":question:")}  {line}'
         return self._get_slack_row(line)
 
     def _get_run_result(self, run: dict) -> dict:
         state = run["status"].get("state", "")
         if state == "error":
-            error_status = run["status"].get("error", "")
+            error_status = run["status"].get("error", "") or state
             result = f"*{error_status}*"
         else:
             result = mlrun.utils.helpers.dict_to_str(
                 run["status"].get("results", {}), ", "
             )
-        return self._get_slack_row(result or "None")
+        return self._get_slack_row(result or state)
 
     @staticmethod
     def _get_slack_row(text: str) -> dict:

mlrun/utils/notifications/notification/webhook.py

@@ -57,7 +57,7 @@ class WebhookNotification(NotificationBase):
             request_body["alert"] = alert.dict()
             if event_data:
                 request_body["value"] = event_data.value_dict
-                request_body["id"] = event_data.entity.id
+                request_body["id"] = event_data.entity.ids[0]
 
         if custom_html:
             request_body["custom_html"] = custom_html

mlrun/utils/notifications/notification_pusher.py

@@ -14,11 +14,17 @@
 
 import asyncio
 import datetime
+import json
 import os
+import re
 import traceback
 import typing
 from concurrent.futures import ThreadPoolExecutor
 
+import kfp
+import mlrun_pipelines.common.ops
+
+import mlrun.common.runtimes.constants
 import mlrun.common.schemas
 import mlrun.config
 import mlrun.db.base
@@ -238,20 +244,7 @@ class NotificationPusher(_NotificationPusherBase):
             custom_message = (
                 f" (workflow: {run.metadata.labels['workflow']}){custom_message}"
             )
-            db = mlrun.get_run_db()
-
-            workflow_id = run.status.results.get("workflow_id", None)
-            if workflow_id:
-                workflow_runs = db.list_runs(
-                    project=run.metadata.project,
-                    labels=f"workflow={workflow_id}",
-                )
-                logger.debug(
-                    "Found workflow runs, extending notification runs",
-                    workflow_id=workflow_id,
-                    workflow_runs_amount=len(workflow_runs),
-                )
-                runs.extend(workflow_runs)
+            runs.extend(self.get_workflow_steps(run))
 
         message = (
             self.messages.get(run.state(), "").format(resource=resource)
@@ -395,6 +388,133 @@ class NotificationPusher(_NotificationPusherBase):
             mask_params=False,
         )
 
+    def get_workflow_steps(self, run: mlrun.model.RunObject) -> list:
+        steps = []
+        db = mlrun.get_run_db()
+
+        def _add_run_step(_node_name, _node_template, _step_kind):
+            _run = db.list_runs(
+                project=run.metadata.project,
+                labels=f"mlrun/runner-pod={_node_name}",
+            )[0]
+            _run["step_kind"] = _step_kind
+            steps.append(_run)
+
+        def _add_deploy_function_step(_, _node_template, _step_kind):
+            project, name, hash_key = self._extract_function_uri(
+                _node_template["metadata"]["annotations"]["mlrun/function-uri"]
+            )
+            if name:
+                try:
+                    function = db.get_function(
+                        project=project, name=name, hash_key=hash_key
+                    )
+                except mlrun.errors.MLRunNotFoundError:
+                    # If the function is not found (if build failed for example), we will create a dummy
+                    # function object for the notification to display the function name
+                    function = {
+                        "metadata": {
+                            "name": name,
+                            "project": project,
+                            "hash_key": hash_key,
+                        },
+                    }
+                function["status"] = {
+                    "state": mlrun.common.runtimes.constants.PodPhases.pod_phase_to_run_state(
+                        node["phase"]
+                    ),
+                }
+                if isinstance(function["metadata"].get("updated"), datetime.datetime):
+                    function["metadata"]["updated"] = function["metadata"][
+                        "updated"
+                    ].isoformat()
+                function["step_kind"] = _step_kind
+                steps.append(function)
+
+        step_methods = {
+            mlrun_pipelines.common.ops.PipelineRunType.run: _add_run_step,
+            mlrun_pipelines.common.ops.PipelineRunType.build: _add_deploy_function_step,
+            mlrun_pipelines.common.ops.PipelineRunType.deploy: _add_deploy_function_step,
+        }
+
+        workflow_id = run.status.results.get("workflow_id", None)
+        if not workflow_id:
+            return steps
+
+        workflow_manifest = self._get_workflow_manifest(workflow_id)
+        if not workflow_manifest:
+            return steps
+
+        try:
+            workflow_nodes = sorted(
+                workflow_manifest["status"]["nodes"].items(),
+                key=lambda _node: _node[1]["finishedAt"],
+            )
+            for node_name, node in workflow_nodes:
+                if node["type"] != "Pod":
+                    # Skip the parent DAG node
+                    continue
+
+                node_template = next(
+                    template
+                    for template in workflow_manifest["spec"]["templates"]
+                    if template["name"] == node["templateName"]
+                )
+                step_type = node_template["metadata"]["annotations"].get(
+                    "mlrun/pipeline-step-type"
+                )
+                step_method = step_methods.get(step_type)
+                if step_method:
+                    step_method(node_name, node_template, step_type)
+            return steps
+        except Exception:
+            # If we fail to read the pipeline steps, we will return the list of runs that have the same workflow id
+            logger.warning(
+                "Failed to extract workflow steps from workflow manifest, "
+                "returning all runs with the workflow id label",
+                workflow_id=workflow_id,
+                traceback=traceback.format_exc(),
+            )
+            return db.list_runs(
+                project=run.metadata.project,
+                labels=f"workflow={workflow_id}",
+            )
+
+    @staticmethod
+    def _get_workflow_manifest(workflow_id: str) -> typing.Optional[dict]:
+        kfp_client = kfp.Client(namespace=mlrun.mlconf.namespace)
+
+        # arbitrary timeout of 5 seconds, the workflow should be done by now
+        kfp_run = kfp_client.wait_for_run_completion(workflow_id, 5)
+        if not kfp_run:
+            return None
+
+        kfp_run = kfp_run.to_dict()
+        try:
+            return json.loads(kfp_run["pipeline_runtime"]["workflow_manifest"])
+        except Exception:
+            return None
+
+    def _extract_function_uri(self, function_uri: str) -> tuple[str, str, str]:
+        """
+        Extract the project, name, and hash key from a function uri.
+        Examples:
+            - "project/name@hash_key" returns project, name, hash_key
+            - "project/name returns" project, name, ""
+        """
+        project, name, hash_key = None, None, None
+        hashed_pattern = r"^(.+)/(.+)@(.+)$"
+        pattern = r"^(.+)/(.+)$"
+        match = re.match(hashed_pattern, function_uri)
+        if match:
+            project, name, hash_key = match.groups()
+        else:
+            match = re.match(pattern, function_uri)
+            if match:
+                project, name = match.groups()
+                hash_key = ""
+        return project, name, hash_key
+
 
 class CustomNotificationPusher(_NotificationPusherBase):
     def __init__(self, notification_types: list[str] = None):

mlrun/utils/version/version.json

@@ -1,4 +1,4 @@
 {
-  "git_commit": "0967b2a28f5a75cc377483f5267159eb75925822",
-  "version": "1.7.0-rc14"
+  "git_commit": "db779a53a32a737e10d538e2fef25695a962117a",
+  "version": "1.7.0-rc16"
 }

{mlrun-1.7.0rc14.dist-info → mlrun-1.7.0rc16.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: mlrun
-Version: 1.7.0rc14
+Version: 1.7.0rc16
 Summary: Tracking and config of machine learning runs
 Home-page: https://github.com/mlrun/mlrun
 Author: Yaron Haviv
@@ -26,7 +26,6 @@ Requires-Dist: GitPython >=3.1.41,~=3.1
 Requires-Dist: aiohttp ~=3.9
 Requires-Dist: aiohttp-retry ~=2.8
 Requires-Dist: click ~=8.1
-Requires-Dist: kfp ~=1.8
 Requires-Dist: nest-asyncio ~=1.0
 Requires-Dist: ipython ~=8.10
 Requires-Dist: nuclio-jupyter ~=0.9.16
@@ -44,13 +43,15 @@ Requires-Dist: semver ~=3.0
 Requires-Dist: dependency-injector ~=4.41
 Requires-Dist: fsspec <2024.4,>=2023.9.2
 Requires-Dist: v3iofs ~=0.1.17
-Requires-Dist: storey ~=1.7.7
+Requires-Dist: storey ~=1.7.11
 Requires-Dist: inflection ~=0.5.0
 Requires-Dist: python-dotenv ~=0.17.0
 Requires-Dist: setuptools ~=69.1
 Requires-Dist: deprecated ~=1.2
 Requires-Dist: jinja2 >=3.1.3,~=3.1
 Requires-Dist: orjson <4,>=3.9.15
+Requires-Dist: mlrun-pipelines-kfp-common-experiment ~=0.2.0
+Requires-Dist: mlrun-pipelines-kfp-v1-8-experiment ~=0.2.0
 Provides-Extra: alibaba-oss
 Requires-Dist: ossfs ==2023.12.0 ; extra == 'alibaba-oss'
 Requires-Dist: oss2 ==2.18.1 ; extra == 'alibaba-oss'