fix: properly consider running buildrun condition (#1006)

Author: olevski

components/renku_data_services/k8s/models.py

@@ -243,13 +243,14 @@ class APIObjectInCluster:
     @property
     def user_id(self) -> str | None:
         """Extract the user id from annotations."""
+        labels = cast(dict[str, str], self.obj.metadata.get("labels", {}))
         match self.obj.singular:
             case "jupyterserver":
-                return cast(str, self.obj.metadata.labels["renku.io/userId"])
+                return labels.get("renku.io/userId", None)
             case "amaltheasession":
-                return cast(str, self.obj.metadata.labels["renku.io/safe-username"])
+                return labels.get("renku.io/safe-username", None)
             case "buildrun":
-                return cast(str, self.obj.metadata.labels["renku.io/safe-username"])
+                return labels.get("renku.io/safe-username", None)
 
             case "taskrun":
                 return DUMMY_TASK_RUN_USER_ID

components/renku_data_services/session/k8s_client.py

@@ -223,8 +223,25 @@ async def update_image_build_status(self, buildrun_name: str, user_id: str) -> m
             return models.ShipwrightBuildStatusUpdate(update=None)
 
         conditions = k8s_build_status.conditions
+        # NOTE: You can get a condition like this in some cases during autoscaling or for other reasons
+        #   message: Not all Steps in the Task have finished executing
+        #   reason: Running
+        #   status: Unknown
+        #   /type: Succeeded
+        # In this case we want to keep waiting - the buildrun is still running.
+        # A fully successful completion condition looks like this:
+        #   reason: Succeeded
+        #   status: True
+        #   /type: Succeeded
+        # See https://shipwright.io/docs/build/buildrun/#understanding-the-state-of-a-buildrun
+        # NOTE: In the examples above I put / before the type field because mypy parses that and fails.
+        # So I needed something to keep mypy happy. The real name of the field is "type"
         condition = next(filter(lambda c: c.type == "Succeeded", conditions or []), None)
 
+        if condition is not None and condition.reason in ["Running", "Pending"]:
+            # The buildrun is still running or pending
+            return models.ShipwrightBuildStatusUpdate(update=None)
+
         buildSpec = k8s_build_status.buildSpec
         output = buildSpec.output if buildSpec else None
         result_image = output.image if output else "unknown"
@@ -238,7 +255,7 @@ async def update_image_build_status(self, buildrun_name: str, user_id: str) -> m
         result_repository_git_commit_sha = git_obj_2.commitSha if git_obj_2 else None
         result_repository_git_commit_sha = result_repository_git_commit_sha or "unknown"
 
-        if condition is not None and condition.status == "True":
+        if condition is not None and condition.reason == "Succeeded" and condition.status == "True":
             return models.ShipwrightBuildStatusUpdate(
                 update=models.ShipwrightBuildStatusUpdateContent(
                     status=models.BuildStatus.succeeded,