|ML-10488|Validation of mail configuration does not fail on non-required parameters. See the updated documentation [Mail notifications](../concepts/notifications.md#mail-notifications).|

|ML-11277|Increased the performance of reading partitioned parquets, thereby preventing lag in processing the model monitoring events.|

|ML-11597|The Model monitoring tutorial now runs successfully. Previously it failed with an error "NameError: name 'alert_objects' is not defined".|

|ML-11791|Documentation: Clarified project.save() and function.save(): functions are not saved when you save your project, and how to save functions to the DB. See [A word about saving functions](../runtimes/create-and-use-functions.ipynb#a-word-about-saving-functions).|

|ML-3854| Support for webhook notification. See webhook in {ref}`notifications` and [view in Git](https://github.com/mlrun/mlrun/pull/3946). |

|ML-11654|MLRun serving graphs with HTTP trigger and no responder. When a serving function is configured with an HTTP trigger only and the graph does not include any `.respond()` step, the function does not return the actual result or error of the graph execution. Instead, it only returns a generic invocation ID (for example, {"id": "<uuid>"}), even if an exception occurred inside the graph.|Add a `.respond()` step.|v1.10.0|

- [SDK](#sdk)

- [MLRun on Iguazio](#mlrun-on-iguazio)

- [MLRun CE](#mlrun-ce)

- {py:class}`~mlrun.common.schemas.notification.Notification`: The notification object

- {py:class}`~mlrun.common.schemas.notification.NotificationKind`: The notification kinds

- {py:class}`~mlrun.db.httpdb.HTTPRunDB.refresh_smtp_configuration`: Gets or refreshes the SMTP configuration from the Iguazio platform and sets it

as the default SMTP configuration (creates an `mlrun-smtp-config` with the SMTP configuration). For privileged user: `IT Admin`.

Inside the notification's `secret_params`, there's a reference to the secret under the `secret` key after it's masked.

When MLRun is deployed on the Iguazio platform, your IT Admin can [configure the SMTP server in the Iguazio platform](https://www.iguazio.com/docs/latest-release/cluster-mgmt/deployment/post-deployment-howtos/smtp/). To use this as your default configuration, run the following (with privileged user - `IT Admin`):

import mlrun

mlrun.get_run_db().refresh_smtp_configuration()

The `refresh_smtp_configuration` method gets the SMTP configuration from the Iguazio platform and sets it

as the default, cluster-wide, SMTP configuration (creates an `mlrun-smtp-config` secret with the SMTP configuration).

If you edit the configuration on the Iguazio platform, run the `refresh_smtp_configuration` method again.

Three parameters cannot be configured on the Iguazio platform. To set their defaults for the cluster, run these commands with the relevant values:

These parameters are maintained when modifying parameters on the Iguazio platform.

In the community edition, you can use your own SMTP server.

To configure it, manually create the `mlrun-smtp-config` Kubernetes secret with the default

parameters for the SMTP server (`server_host`, `server_port`, `username`, `password`, `start_tls`, etc.).

After creating or editing the secret, refresh the MLRun SMTP configuration by running the `refresh_smtp_configuration` method.

The following snippet shows the format of a notification object.

You can inherit the default `mlrun-smtp-config`, or choose to overwrite parameter/s.

Any `params` not defined in this format will be enriched with the values in the `mlrun-smtp-config` secret.

The only mandatory field in `params` is `email_addresses`.

"```{admonition} Note\n",

"When using an application runtime with min-replicas set to 0 (Scale-To-Zero enabled) and direct port access enabled , be aware of a known limitation in metric collection. Scale-to-Zero currently relies on metrics collected only from the main Nuclio container that runs the reverse proxy. Traffic sent directly to the application runtime sidecar bypasses these metrics, so such requests are not counted. As a result, the function may scale down to zero even while it is actively serving traffic through the direct port. In addition, once scaled to zero, the function can only be scaled back up by invoking the main Nuclio port, not the direct port. To avoid unexpected behavior, do not rely on direct port traffic alone when using Scale-To-Zero.\n",

"```\n",

"\n",