[Yandex Cloud documentation](../../../index.md) > [Yandex Compute Cloud](../../index.md) > [Step-by-step guides](../index.md) > Managing a VM > Managing maintenance of GPU VMs # Managing maintenance of GPU VMs Compute Cloud performs occasional [maintenance](../../concepts/gpus-maintenance.md) of GPU VMs and notifies users of the scheduled date and time. You can view the list of maintenance [tasks](../../concepts/gpus-maintenance.md#tasks) scheduled for your VMs and reschedule them as needed. ## Viewing VM maintenance tasks {% list tabs group=instructions %} - Management console {#console} 1. In the [management console](https://console.yandex.cloud), select the [folder](../../../resource-manager/concepts/resources-hierarchy.md#folder) where you want to view VM maintenance tasks. 1. Navigate to **Compute Cloud**. 1. In the left-hand panel, select ![bars-play](../../../_assets/console-icons/bars-play.svg) **Maintenance**. The window that opens displays a list of all maintenance tasks for GPU VMs, which includes: * `ID`: Task ID. * `Task`: Action performed during maintenance affecting the VM availability, e.g., `Restart`. * `Status`: Task [status](../../concepts/gpus-maintenance.md#maintenance-status). * `Resource`: Name of the VM to undergo maintenance. * `Start date`: Start date and time of the scheduled VM maintenance if the task is pending; otherwise, the actual maintenance start date and time. * `Completed`: Maintenance completion date and time. To view [details](../../concepts/gpus-maintenance.md#tasks) of a maintenance task, click the relevant row in the task list. Use the top-of-page filter to quickly find the task you need in the list. Dates and times displayed in the task details match the time zone selected in the user’s [management console](https://console.yandex.cloud) settings. {% note tip %} You can also view details of a scheduled VM maintenance task in a note ![bulb](../../../_assets/console-icons/bulb.svg) within the **Name** field in the list of VMs under **Virtual machines** and on the relevant VM page. {% endnote %} - CLI {#cli} If you do not have the Yandex Cloud CLI yet, [install and initialize it](../../../cli/quickstart.md#install). The folder used by default is the one specified when [creating](../../../cli/operations/profile/profile-create.md) the CLI profile. To change the default folder, use the `yc config set folder-id ` command. You can also specify a different folder for any command using `--folder-name` or `--folder-id`. If you access a resource by its name, the search will be limited to the default folder. If you access a resource by its ID, the search will be global, i.e., through all folders based on access permissions. 1. To view the list of VM maintenance tasks, run the following command: ```bash yc compute maintenance list \ --cloud-id \ --folder-id \ --resource-id \ --filter \ --order ``` Where: * `--cloud-id`: [ID](../../../resource-manager/operations/cloud/get-id.md) of the cloud where you want to view tasks. * `--folder-id`: [ID](../../../resource-manager/operations/folder/get-id.md) of the folder where you want to view tasks. * `--resource-id`: ID of the VM to view tasks for. {% note info %} `--cloud-id`, `--folder-id`, and `--resource-id` are mutually exclusive: you can use only one of them. If none of these parameters is set, the command will output the list of maintenance tasks in the default [folder](../../../resource-manager/concepts/resources-hierarchy.md#folder). {% endnote %} * `--filter`: Filter to apply to the task list you obtained. This is an optional setting. By default, the command outputs the full list of maintenance tasks. You can filter the list by task status (`status`) or by VM ID (`resource_id`), e.g., `--filter 'status = "SCHEDULED"'` or `--filter 'resource_id = "a7loutvf73vg********"'`. * `sort_order`: Task list sorting order. This is an optional setting. Available values: * `created_at`: By the task creation time. * `start_scheduled_at`: By the scheduled time of the task. * `started_at`: By the task start time. By default, tasks are sorted in ascending order. To sort tasks in descending order, use the `DESC` setting, e.g., `--order 'created_at DESC'`. Result: ```bash +----------------------+----------------------+----------------------+------------------+----------------------+-----------+---------------------+---------------------+---------------------+ | ID | CLOUD ID | FOLDER ID | RESOURCE TYPE | RESOURCE ID | STATUS | CREATED AT | START SCHEDULED AT | STARTED AT | +----------------------+----------------------+----------------------+------------------+----------------------+-----------+---------------------+---------------------+---------------------+ | a7l0b4nkle3b******** | a7lia87mbaom******** | a7lt6g8ht345******** | compute.instance | a7loutvf73vg******** | SUCCEEDED | 2025-09-03 11:08:15 | 2025-09-03 11:15:39 | 2025-09-03 11:15:52 | | a7l742uljpl3******** | a7lia87mbaom******** | a7lt6g8ht345******** | compute.instance | a7l7f7n9dbfq******** | SCHEDULED | 2025-09-03 11:58:59 | 2025-09-08 07:00:00 | | +----------------------+----------------------+----------------------+------------------+----------------------+-----------+---------------------+---------------------+---------------------+ ``` The `ID` field contains the task ID. 1. To view information about a specific task, run the following command with the ID of your task: ```bash yc compute maintenance get ``` Result: ```bash id: a7l742uljpl3******** resource_path: - id: a7l7f7n9dbfq******** type: compute.instance - id: aoepotm5k3oa******** type: resource-manager.folder - id: a7lia87mbaom******** type: resource-manager.cloud status: SCHEDULED created_at: "2025-09-03T11:58:59Z" start_scheduled_at: "2025-09-08T07:00:00Z" completion_scheduled_at: "2025-09-08T15:00:00Z" max_start_scheduled_at: "2025-09-08T07:00:00Z" user_controllable: true ``` {% endlist %} ## Rescheduling VM maintenance tasks {#reschedule-maintenance} You can reschedule a task to an earlier or later date and time: {% list tabs group=instructions %} - Management console {#console} 1. In the [management console](https://console.yandex.cloud), select the [folder](../../../resource-manager/concepts/resources-hierarchy.md#folder) containing the VM to reschedule a maintenance task for. 1. Navigate to **Compute Cloud**. 1. In the left-hand panel, select ![bars-play](../../../_assets/console-icons/bars-play.svg) **Maintenance**. 1. In the list that opens, click ![ellipsis](../../../_assets/console-icons/ellipsis.svg) next to the task in question and select ![arrow-uturn-cw-right](../../../_assets/console-icons/arrow-uturn-cw-right.svg) **Reschedule**. In the window that opens: 1. Select a new date and time window for the task. To see how far you can postpone the maintenance task, click the task row in the list. In the window that opens, the **Can be postponed until** field shows the latest possible date and time. 1. Click **Reschedule**. {% note tip %} You can also reschedule a VM maintenance task by clicking ![bulb](../../../_assets/console-icons/bulb.svg) in the **Name** field in the VM list under **Virtual machines**, and on the relevant VM page. {% endnote %} - CLI {#cli} If you do not have the Yandex Cloud CLI yet, [install and initialize it](../../../cli/quickstart.md#install). The folder used by default is the one specified when [creating](../../../cli/operations/profile/profile-create.md) the CLI profile. To change the default folder, use the `yc config set folder-id ` command. You can also specify a different folder for any command using `--folder-name` or `--folder-id`. If you access a resource by its name, the search will be limited to the default folder. If you access a resource by its ID, the search will be global, i.e., through all folders based on access permissions. 1. [Get](#list-scheduled-maintenance) the ID of the maintenance task you need. 1. To reschedule the VM maintenance task, run this command: ```bash yc compute maintenance reschedule \ --time "" ``` Where: * ``: ID of the task to reschedule, obtained in the previous step. * ``: New date and time for the task, e.g., `2025-09-07T07:00:00Z`. Provide the time in [ISO 8601](https://en.wikipedia.org/wiki/ISO_8601) UTC format with zero offset. For Moscow time, add `+03:00` instead of `Z` at the end of the line: `2025-09-07T07:00:00+03:00`. Result: ```bash id: a7l742uljpl3******** resource_path: - id: a7l7f7n9dbfq******** type: compute.instance - id: aoepotm5k3oa******** type: resource-manager.folder - id: a7lia87mbaom******** type: resource-manager.cloud status: SCHEDULED created_at: "2025-09-03T11:58:59Z" start_scheduled_at: "2025-09-07T07:00:00Z" completion_scheduled_at: "2025-09-07T15:00:00Z" max_start_scheduled_at: "2025-09-08T07:00:00Z" user_controllable: true ``` {% endlist %} ## Instant start of scheduled maintenance tasks {#execute-now} You can instantly start a scheduled VM maintenance task. {% list tabs group=instructions %} - Management console {#console} 1. In the [management console](https://console.yandex.cloud), select the [folder](../../../resource-manager/concepts/resources-hierarchy.md#folder) containing the VM to start a maintenance task for. 1. Navigate to **Compute Cloud**. 1. In the left-hand panel, select ![bars-play](../../../_assets/console-icons/bars-play.svg) **Maintenance**. 1. In the list that opens, click ![ellipsis](../../../_assets/console-icons/ellipsis.svg) next to the task in question and select ![play](../../../_assets/console-icons/play.svg) **Perform now**. 1. In the window that opens, confirm that you want to instantly start the task. {% note tip %} You can also instantly start a VM maintenance task by clicking ![bulb](../../../_assets/console-icons/bulb.svg) in the **Name** field in the VM list under **Virtual machines**, and on the relevant VM page. {% endnote %} - CLI {#cli} If you do not have the Yandex Cloud CLI yet, [install and initialize it](../../../cli/quickstart.md#install). The folder used by default is the one specified when [creating](../../../cli/operations/profile/profile-create.md) the CLI profile. To change the default folder, use the `yc config set folder-id ` command. You can also specify a different folder for any command using `--folder-name` or `--folder-id`. If you access a resource by its name, the search will be limited to the default folder. If you access a resource by its ID, the search will be global, i.e., through all folders based on access permissions. 1. [Get](#list-scheduled-maintenance) the ID of the maintenance task you need. 1. To instantly start a task, run the following command with the task ID you obtained in the previous step: ```bash yc compute maintenance reschedule \ --immediate ``` Result: ```bash id: a7l742uljpl3******** resource_path: - id: a7l7f7n9dbfq******** type: compute.instance - id: aoepotm5k3oa******** type: resource-manager.folder - id: a7lia87mbaom******** type: resource-manager.cloud status: RUNNING created_at: "2025-09-03T11:58:59Z" start_scheduled_at: "2025-09-03T13:07:13.136955876Z" completion_scheduled_at: "2025-09-04T21:07:13.136955876Z" max_start_scheduled_at: "2025-09-08T07:00:00Z" user_controllable: true ``` {% endlist %} #### See also {#see-also} * [Maintenance of GPU VMs](../../concepts/gpus-maintenance.md)