Manage host events across reservations

This document explains how to use the host maintenance features that are available with your reserved blocks of capacity. It explains how to monitor, plan for, and perform scheduled maintenance on your cluster of H4D instances. To manage maintenance on your virtual machine (VM) instances instead, see Monitor and plan for a host maintenance event.

You can proactively manage upcoming maintenance host events on your reserved blocks of capacity, whether VMs are running on them or not. This approach helps you minimize disruptions and maintain optimal performance.

Before you begin

If you haven't already, set up authentication. Authentication verifies your identity for access to Google Cloud services and APIs. To run code or samples from a local development environment, you can authenticate to Compute Engine by selecting one of the following options:
Select the tab for how you plan to use the samples on this page:
Console

When you use the Google Cloud console to access Google Cloud services and APIs, you don't need to set up authentication.
gcloud
1. Install the Google Cloud CLI. After installation, initialize the Google Cloud CLI by running the following command:
  gcloud init
  If you're using an external identity provider (IdP), you must first sign in to the gcloud CLI with your federated identity.
  
  Note: If you installed the gcloud CLI previously, make sure you have the latest version by running gcloud components update.
2. Set a default region and zone.
REST

To use the REST API samples on this page in a local development environment, you use the credentials you provide to the gcloud CLI.
For more information, see Authenticate for using REST in the Google Cloud authentication documentation.

Required roles

To get the permissions that you need to manage host maintenance events across reservations, ask your administrator to grant you the following IAM roles:

Compute Admin (roles/compute.admin) on the project
For read-only access to System Event audit logs: Logs Viewer (roles/logging.viewer) on the project

For more information about granting roles, see Manage access to projects, folders, and organizations.

These predefined roles contain the permissions required to manage host maintenance events across reservations. To see the exact permissions that are required, expand the Required permissions section:

Required permissions

The following permissions are required to manage host maintenance events across reservations:

To start host maintenance for a reservation: compute.reservations.performMaintenance on the project

To start host maintenance for a reservation block: compute.reservationBlocks.performMaintenance on the project

To view a list reservations

You might also be able to get these permissions with custom roles or other predefined roles.

Overview

To optimize the maintenance of your reserved blocks of capacity, complete the following steps:

Set up notification alerts. Create log-based alerts to get notified about scheduled, started, or completed maintenance events for a reservation or a reservation block. This approach helps you proactively plan your activities and avoid unexpected downtime.
Manage maintenance across blocks of capacity. View and, if needed, manually start maintenance across your reservations or reservation blocks. This process helps you increase the resilience of your workloads to host errors, prevent downtime, and ensure that your applications remain available.

For more information about the frequency and maintenance behavior of H4D machine types, see Maintenance experience for H4D instances.

Set up notification alerts for reservations

You can get notified about scheduled, started, or completed maintenance events for a reservation or reservation block by creating log-based alerting policies.

To create an alert for the maintenance events of a reservation or a reservation block, complete the following procedure. Repeat this procedure for each alert that you want to create.

In the Google Cloud console, go to the Logs Explorer page:
Go to Logs Explorer

If you use the search bar to find this page, then select the result whose subheading is Logging.
Click the Show query toggle to the on position.

In the Query pane, build one of the following queries. These queries filter log entries to identify specific maintenance events. Repeat this procedure for each query you want to create.

Receive maintenance alerts for a reservation:

To receive alerts when maintenance is scheduled:

protoPayload.methodName="compute.reservations.upcomingGroupMaintenance" severity>=DEFAULT
protoPayload.status.message =~ "scheduled"

To receive alerts when maintenance has completed:

protoPayload.methodName="compute.reservations.completedGroupMaintenance" severity>=DEFAULT
protoPayload.status.message =~ "completed"

Receive maintenance alerts for a reservation block:

To receive alerts when maintenance is scheduled:

protoPayload.methodName="compute.reservations.block.upcomingGroupMaintenance" severity>=DEFAULT
protoPayload.status.message =~ "scheduled"

To receive alerts when maintenance has started:

protoPayload.methodName="compute.reservations.block.startGroupMaintenance" severity>=DEFAULT
protoPayload.status.message =~ "started"

To receive alerts when maintenance has completed:

protoPayload.methodName="compute.reservations.block.completedGroupMaintenance" severity>=DEFAULT
protoPayload.status.message =~ "completed"

To validate the query, click Run query. If the query is valid, then the Query results pane displays log entries that match the query.
In the Query results toolbar, click the Actions list, and then select Create log alert. The Create logs-based alert policy pane appears.
In the Alert details section, do the following:
1. In the Alert Policy Name field, enter a name for the policy.
2. In the Policy severity level list, select Warning (or a higher severity).
3. Click Next.
In the Choose logs to include in the alert section, click Next.
In the Set notification frequency and autoclose duration section, specify the following:
1. In the Time between notifications list, select how often you want to be notified.
2. In the Incident autoclose duration list, select after how long Cloud Logging stops sending notifications and automatically closes the incident.
3. Click Next.
In the Who should be notified? section, specify a notification channel for Logging to send notifications to.
Click Save.

Manage maintenance across reservations

You can view and control maintenance for your reservations and reservation blocks as follows:

To check the state and scheduled time of upcoming maintenance for your reservations or reservation blocks, view maintenance state.
To manually start maintenance on a reservation or reservation block, rather than waiting for the scheduled maintenance date and time, manually start maintenance.
To manage how early you want to receive notifications when a VM's host requires emergency, unplanned maintenance after a host error or faulty host report, manage hardware emergency maintenance notifications.

View maintenance state

You can view the upcoming maintenance state for a reservation or a reservation block by checking the value of the upcomingGroupMaintenance field in the reservation metadata. If a reservation lacks the upcomingGroupMaintenance field, then no maintenance is scheduled for the reservation or reservation block. For more information about the fields in upcomingGroupMaintenance, see Maintenance status definitions.

Additionally, if maintenance is scheduled for a reservation block, the upcomingGroupMaintenance field contains the maintenanceReasons field. This field describes why maintenance was scheduled for your reservation block, as described in the following table:

Maintenance type	Maintenance reason	VMs state
Planned maintenance for regular maintenance	`PLANNED_UPDATE`: planned infrastructure maintenance `PLANNED_NETWORK_UPDATE`: planned network maintenance	Applies to running, stopped, or suspended VMs.
Unplanned, emergency maintenance	`FAILURE_DISK`: unplanned maintenance scheduled due to an error with a disk attached to a VM. `FAILURE_INFRA`: unplanned maintenance scheduled due to an error with the infrastructure. `FAILURE_INTERFACE`: unplanned maintenance scheduled due to an error with the NIC. `FAILURE_MEMORY`: unplanned maintenance scheduled due to an error with a VM's memory. `FAILURE_NETWORK`: unplanned maintenance scheduled due to an error with the network.	Applies only to running VMs.

To view the maintenanceReasons field in a reservation block or sub-block, you must use the gcloud CLI or REST API. Otherwise, select one of the following options:

Console

In the Google Cloud console, go to the Reservations page.

Go to Reservations
In the Maintenance status column, Compute Engine displays the maintenance state of your reservations. If you don't see this column in the On-demand reservations table, then click Column display options, select the Maintenance status checkbox, and then click OK.
To view the maintenance state of a reservation block, complete the following steps:
1. In the Name column, click the name of the reservation. A page that gives the details of the reservation appears.
2. In the Blocks table, in the Maintenance column, Compute Engine displays the maintenance state of the blocks within the reservation.

gcloud

To view the maintenance state of a reservation, use the gcloud compute reservations describe command with the --flatten flag set to resourceStatus.reservationMaintenance:
```
gcloud compute reservations describe RESERVATION_NAME \
    --flatten=resourceStatus.reservationMaintenance \
    --zone=ZONE
```
Replace the following:
- RESERVATION_NAME: the name of the reservation.
- ZONE: the zone where the reservation exists.
The output is similar to one of the following:
- If maintenance is scheduled for your reservation, then the output is similar to the following:
```
---
maintenanceOngoingCount: 0
maintenancePendingCount: 6
schedulingType: GROUPED
upcomingGroupMaintenance:
  canReschedule: true
  maintenanceStatus: PENDING
  type: UNSCHEDULED
  windowEndTime: '2025-11-13T14:00:00.000-08:00'
  windowStartTime: '2025-11-13T12:00:00.000-08:00'
```
  If the schedulingType field is set to INDEPENDENT, then the upcomingGroupMaintenance field doesn't contain the windowStartTime and windowEndTime fields.
- If maintenance isn't scheduled for your reservation, then the output is similar to the following:
```
---
schedulingType: GROUPED
```

If maintenance is scheduled for your reservation, then, to view the maintenance state of the blocks within the reservation, use the gcloud compute reservations blocks list command:

gcloud compute reservations blocks list RESERVATION_NAME \
    --zone=ZONE

If maintenance is scheduled or ongoing for a reservation block, then the output is similar to the following:

---
...
name: example-fr-h4d-dense-1-block-0001
...
reservationBlockMaintenance:
  maintenanceOngoingCount: 0
  maintenancePendingCount: 6
  schedulingType: GROUPED
  upcomingGroupMaintenance:
    canReschedule: true
    maintenanceReasons:
    - PLANNED_UPDATE
    - PLANNED_NETWORK_UPDATE
    maintenanceStatus: PENDING
    type: UNSCHEDULED
    windowEndTime: '2025-11-13T14:00:00.000-08:00'
    windowStartTime: '2025-11-13T12:00:00.000-08:00'
...
---
...
name: example-fr-h4d-dense-1-block-0002
...
schedulingType: GROUPED
...

REST

To view the maintenance state of your reservations, make a GET request to one of the following methods:
- To view reservations across all zones, use the reservations.aggregatedList method.
- To view reservations in a specific zone, use the reservations.list method.
In the request URL, include the following query parameters:
- To only show the name, reserved machine type, and maintenance status of a reservation, include the fields query parameter set to items.name,items.specificReservation.instanceProperties.machineType,items.resourceStatus.reservationMaintenance.
- To only filter by reservations that specify a specific machine type, include the filter query parameter set to specificReservation.instanceProperties.machineType:MACHINE_TYPE by using URL-encoded values.
For example, to view reservations across all zones, make a GET request as follows:
```
GET https://compute.googleapis.com/compute/v1/projects/PROJECT_ID/aggregated/reservations?fields=items.name,items.specificReservation.instanceProperties.machineType,items.resourceStatus.reservationMaintenance&filter=specificReservation.instanceProperties.machineType%3AMACHINE_TYPE
```
Replace the following:
- PROJECT_ID: the ID of the project where the reservations exist.
- MACHINE_TYPE: the reserved machine type that you want to filter your reservations by.
If maintenance is scheduled or ongoing for a reservation, then the output is similar to the following:
```
{
  "items": [
    {
      "specificReservation": {
        "instanceProperties": {
          "machineType": "MACHINE_TYPE"
        }
      },
      "name": "example-reservation",
      "resourceStatus": {
        "reservationMaintenance": {
          maintenanceOngoingCount: 0,
          maintenancePendingCount: 6,
          "schedulingType": "GROUPED",
          "upcomingGroupMaintenance": {
            "type": "SCHEDULED",
            "canReschedule": true,
            "windowStartTime": "2025-11-13T12:00:00.000-08:00",
            "windowEndTime": "2025-11-13T14:00:00.000-08:00",
            "maintenanceStatus": "PENDING"
          }
        }
      }
    },
    ...
  ]
}
```
Optionally, to further narrow down a list of VMs, set the filter query parameter to a different filter expression.

If maintenance is scheduled for your reservation, then, to view the maintenance state of the blocks within the reservation, make a GET request to the reservationBlocks.list method. In the request URL, include the fields query parameter set to items.name,items.reservationMaintenance:

GET https://compute.googleapis.com/compute/v1/projects/PROJECT_ID/zones/ZONE/reservations/RESERVATION_NAME/reservationBlocks?fields=items.name,items.reservationMaintenance

Replace RESERVATION_NAME with the name of the reservation.

If maintenance is scheduled or ongoing for a reservation block, then the output is similar to the following:

{
  "items": [
    {
      "name": "example-fr-h4d-dense-1-block-0001",
      "reservationBlockMaintenance": {
        "maintenanceOngoingCount": 0,
        "maintenancePendingCount": 6,
        "schedulingType": "GROUPED",
        "upcomingGroupMaintenance": {
          "type": "SCHEDULED",
          "canReschedule": true,
          "windowStartTime": "2025-11-13T12:00:00.000-08:00",
          "windowEndTime": "2025-11-13T14:00:00.000-08:00",
          "maintenanceStatus": "PENDING",
          "maintenanceReasons": [
            "PLANNED_UPDATE",
            "PLANNED_NETWORK_UPDATE"
          ]
        }
      }
    },
    ...
  ]
}

If the schedulingType field is set to INDEPENDENT for a block, then the upcomingGroupMaintenance field doesn't contain the windowStartTime and windowEndTime fields.

Manually start maintenance

You can manually start maintenance for your reservations or reservation blocks instead of waiting for the scheduled time. This action helps you more proactively control disruptions to your workloads.

Depending on the maintenance state of a reservation or reservation block, the following occurs:

Maintenance state	Description	What you see
Scheduled	Compute Engine has scheduled maintenance for the reservation. You can manually start maintenance before the scheduled time.	In the Google Cloud console, the maintenance state shows as Ready to run - will run on `DATE`. In the gcloud CLI or REST API, Compute Engine sets the `maintenanceStatus` field to `PENDING`.
In progress	Maintenance is underway. You can't reschedule it.	In the Google Cloud console, the maintenance state shows as Running. In the gcloud CLI or REST API, Compute Engine sets the `maintenanceStatus` field to `ONGOING`.
Complete	Maintenance is finished. Compute Engine has removed all maintenance notifications from the VM.	In the Google Cloud console, the maintenance state shows as —. In the gcloud CLI or REST API, the `maintenanceStatus` field doesn't exist.

To manually start maintenance on specific hosts within a reservation block, use the gcloud CLI or REST API. Otherwise, select one of the following options:

Console

In the Google Cloud console, go to the Reservations page.

Go to Reservations
In the Name column, click the name of a reservation. A page that gives the details of the reservation appears.
Click Run maintenance, and then select one of the following options:
- To start maintenance on all blocks, select All capacity.
- To start maintenance only on blocks with running VMs, select In-use capacity.
- To start maintenance only on unused blocks and blocks with stopped or suspended VMs, select Unused capacity.
To confirm, click Ok.

gcloud

To start maintenance on a reservation, use the gcloud compute reservations perform-maintenance command:

gcloud compute reservations perform-maintenance RESERVATION_NAME \
    --scope=RESERVATION_MAINTENANCE_SCOPE \
    --zone=ZONE

To start maintenance on a reservation block, use the gcloud compute reservations blocks perform-maintenance command:

gcloud compute reservations blocks perform-maintenance RESERVATION_NAME \
    --block-name=BLOCK_NAME \
    --scope=BLOCK_MAINTENANCE_SCOPE \
    --zone=ZONE

Replace the following:

RESERVATION_NAME: the name of the reservation.
RESERVATION_MAINTENANCE_SCOPE: the maintenance scope for the reservation. Specify one of the following values:
- To start maintenance on all blocks: all
- To start maintenance only on blocks with running VMs: running
- To start maintenance only on unused blocks and blocks with stopped or suspended VMs: unused
BLOCK_NAME: the name of a block that exists within the reservation.
BLOCK_MAINTENANCE_SCOPE: the maintenance scope for the reservation block. Specify one of the following values:
- To start maintenance on all hosts: all
- To start maintenance only on hosts with running VMs: running
- To start maintenance only on unused hosts and hosts with stopped or suspended VMs: unused
ZONE: the zone where the reservation exists.

REST

To start maintenance on a reservation, make a POST request to the reservations.performMaintenance method:

POST https://compute.googleapis.com/compute/v1/projects/PROJECT_ID/zones/ZONE/reservations/RESERVATION_NAME/performMaintenance

{
  "maintenanceScope": "RESERVATION_MAINTENANCE_SCOPE"
}

To start maintenance on a reservation block, make a POST request to the reservationBlocks.performMaintenance method:

POST https://compute.googleapis.com/compute/v1/projects/PROJECT_ID/zones/ZONE/reservations/RESERVATION_NAME/reservationBlocks/BLOCK_NAME/performMaintenance

{
  "maintenanceScope": "BLOCK_MAINTENANCE_SCOPE"
}

Replace the following:

PROJECT_ID: the ID of the project where Compute Engine automatically created the reservation.
ZONE: the zone where the reservation exists.
RESERVATION_NAME: the name of the reservation.
RESERVATION_MAINTENANCE_SCOPE: the maintenance scope for the reservation. Specify one of the following values:
- To start maintenance on all blocks: ALL
- To start maintenance only on blocks with running VMs: RUNNING
- To start maintenance only on unused blocks and blocks with stopped or suspended VMs: UNUSED
BLOCK_NAME: the name of a block that exists within the reservation.
SUB_BLOCK_NAME: the name of a sub-block that exists within the reservation block.
BLOCK_MAINTENANCE_SCOPE: the maintenance scope for the reservation block. Specify one of the following values:
- To start maintenance on all hosts: ALL
- To start maintenance only on hosts with running VMs: RUNNING
- To start maintenance only on unused hosts and hosts with stopped or suspended VMs: UNUSED

Manage hardware emergency maintenance notifications

After a VM encounters an host error, or you report its host as faulty, the VM's host requires emergency, unplanned maintenance. By default, Compute Engine provides a few hours of advance notice when it schedules this type of maintenance. For reserved hosts, you can enable emergency maintenance notifications to increase this notice period to at least seven days. This approach helps you more proactively control disruptions to your workloads.

To enable or disable hardware emergency maintenance notifications for a reservation, select one of the following options:

gcloud

To enable hardware emergency maintenance notifications for a reservation, use the gcloud compute reservations update command with the --enable-emergent-maintenance flag:
```
gcloud compute reservations update RESERVATION_NAME \
    --enable-emergent-maintenance \
    --zone=ZONE
```
To disable hardware emergency maintenance notifications for a reservation, use the gcloud compute reservations update command with the --no-enable-emergent-maintenance flag:
```
gcloud compute reservations update RESERVATION_NAME \
    --no-enable-emergent-maintenance \
    --zone=ZONE
```

Replace the following:

RESERVATION_NAME: the name of the reservation.
ZONE: the zone where the reservation exists.

REST

To enable or disable hardware emergency maintenance notifications for a reservation, make a PATCH request to the reservations.update method. In the request URL, include the paths query parameter set to enableEmergentMaintenance.

PATCH https://compute.googleapis.com/compute/v1/projects/PROJECT_ID/zones/ZONE/reservations/RESERVATION_NAME?paths=enableEmergentMaintenance

{
  "name": "RESERVATION_NAME",
  "enableEmergentMaintenance": EMERGENCY_MAINTENANCE_NOTIFICATIONS
}

Replace the following:

PROJECT_ID: the ID of the project where the reservation exists.
ZONE: the zone where the reservation exists.
RESERVATION_NAME: the name of the reservation.
EMERGENCY_MAINTENANCE_NOTIFICATIONS: specify one of the following values:
- To enable notifications: true
- To disable notifications: false

What's next

Learn about Local SSD data persistence during host events
Learn how to perform Local SSD data backup to preserve data
Learn about Maintenance status definitions when monitoring host maintenance events