Skip to main content

Create and manage parameter groups

Parameter groups allow you to view and manage the parameters required for MemStore Redis OSS configuration according to the engine version in one place.
You can also manage the parameters of clusters connected to the parameter group collectively.

The parameter list and basic concepts provided by MemStore in KakaoCloud are compatible with Redis OSS. Changing parameters may affect the MemStore service, so it is recommended to review the appropriate parameters to be entered in advance.

Below is how to manage parameter groups in the MemStore service.

info

View parameter group list​

You can view the list of parameter groups, including default and custom groups, provided by KakaoCloud MemStore.

  1. Go to KakaoCloud Console > Data Store > MemStore menu.

  2. Click the Parameter Group menu and select the Custom group/Default group tab according to the type of parameter group to view.

  3. Navigate to the Parameter Group menu that matches the selected tab.

    ItemDescription
    NameName of the parameter group
    Cluster modeIndicates whether cluster mode is enabled
    FamilyMemStore family engine version compatible with the parameter group
    Number of connected clustersNumber of clusters connected to this group
    - A warning icon is displayed if there is an issue with the parameter application status of the connected cluster
    DescriptionDescription of the parameter group
    MoreAvailable only for custom groups
    - Modify: Modify parameters
    - Copy: Create a new parameter group with the same parameters as this group
    - Reset to default: Reset the group's parameters to the initial values set at the time of creation
    - Rollback: Revert the parameter group to the previous values. Rollback is not possible if there are no modifications or if it has already been rolled back
    - Delete: Delete the parameter group

View parameter group details​

You can view detailed information about the MemStore parameter group.

  1. Go to KakaoCloud Console > Data Store > MemStore menu.

  2. Click the Parameter Group menu and select the Custom group/Default group tab according to the type of parameter group to view.

  3. Navigate to the Parameter Group menu that matches the selected tab and select the parameter group to view detailed information.

    ItemDescription
    NameParameter name
    ValueValue of the parameter
    TypeType of the parameter
    - Some Dynamic parameters require session refresh for global application
    Data formatData format of the parameter
    Editable statusNot editable, Editable, Editable (Modified)
    - Only Editable parameters can be modified, and modified parameters are displayed as Editable (Modified)

Create parameter group​

  1. Go to KakaoCloud Console > Data Store > MemStore menu.

  2. Click the Parameter Group menu and select the Custom group tab.

  3. In the Custom group tab, click the icon > Create Parameter Group.

  4. In the Create Parameter Group page, enter the required information. Parameter values can be modified after the parameter group is created.

    ItemCategoryDescription
    Basic settingsParameter group nameName of the parameter group
    - The same name cannot be used within the same project
    Description (optional)Additional description for the parameter group, up to 100 characters
    Parameter group settingsFamilyMemStore family engine version of the parameter group
    Cluster modeIndicates whether cluster mode is enabled
    Default parameter groupSelect the default parameter group provided for that engine version
    ParametersInitial parameter values of the parameter group

Modify parameter group​

Custom parameter groups can be modified after the group is created. Only parameters with Editable status can be modified. Modified parameters are displayed as Editable (Modified).

info
  • Some Dynamic parameters require session refresh for global application.
  • If the parameter application status of a connected cluster is Pending or Applying, you cannot modify the parameters.
  • In the parameter group modification page, the selected checkboxes remain even if you navigate through the list of editable parameters.
caution
  • Modifying certain parameters may affect MemStore service operations. For more details, refer to Parameters to be careful when modifying.
  • If issues arise after modifying parameters, it is recommended to copy the parameter group for backup and then proceed with Retry/Rollback/Reset.

  1. Go to KakaoCloud Console > Data Store > MemStore menu.

  2. Click the Parameter Group menu and select the Custom group tab.

  3. Select the parameter group to be modified, then click the [More] icon > Modify.

  4. Modify the necessary information in the Modify Parameter Group page.

    ItemCategoryDescription
    Basic informationParameter group nameName of the parameter group
    Description (optional)Additional description for the parameter group, up to 100 characters
    Editable parametersList of editable parameters within the parameter group
    - Modified parameters are displayed as Editable (Modified)

  5. If multiple clusters are connected to the parameter group, you can select the parameter application method.
    Apply all: Apply the modified parameters to all clusters simultaneously.
    Sequential apply: Apply the modified parameters sequentially to clusters.

Parameters to be careful when modifying​

When modifying parameter values in a custom parameter group, modifying certain parameters may affect MemStore service operations. Please refer to the following details for more information.

info

repl-backlog

  • The buffer temporarily stores data to be sent to the replica node for partial synchronization if the primary and replica nodes are disconnected due to a network interruption.
  • If a network interruption occurs, the primary node accumulates data for synchronization in the backlog. When the replica reconnects, partial synchronization (psync) is attempted using the data stored in the backlog.
  • The primary node maintains backlog data only for a certain period (repl-backlog-ttl) after the last disconnection from the replica node, after which the backlog buffer is released. After the backlog buffer is released, if the replica node reconnects, a full sync is performed for data synchronization.
  • Related parameters: repl-backlog-size, repl-backlog-ttl

client-output-buffer-limit-normal-hard-limit​

TypeInput rangeDescriptionPossible issues
integer0~Upper limit of client output buffer memory usage
Setting to 0 means no limit.		If the client output buffer memory reaches this parameter value, the client connection is disconnected.
Setting the value too low may cause the MemStore service to fail to provision.
It is recommended to set a sufficient value or the default value (0) considering the workload.

물론이죠! 아래와 같이 마크다운 코드 블록을 사용해 영문 내용을 제공할 수 있습니다:

### client-output-buffer-limit-normal-soft-limit

| Type    | Input range | Description                                                                                                          |
|---------|-------------|----------------------------------------------------------------------------------------------------------------------|
| integer | 0~          | The threshold for client output buffer memory usage <br/> Allowed for a certain period (client-output-buffer-limit-normal-soft-seconds). <br/> Setting to `0` means no threshold. <br/><br/> **Possible issues**: <br/> If the client output buffer memory usage exceeds this parameter value for longer than the threshold time, the client connection is disconnected. <br/> Setting the value too low may cause the MemStore service to fail to provision. <br/> It is recommended to set a sufficient value or the default value (0) considering the workload. |

### client-output-buffer-limit-normal-soft-seconds

| Type    | Input range  | Description                                                                                                          |
|---------|--------------|----------------------------------------------------------------------------------------------------------------------|
| integer | 0~2147483647 | The time allowed for the client output buffer to exceed the set threshold (seconds). <br/> Setting to `0` means no time limit. <br/><br/> **Possible issues**: <br/> If the client output buffer memory usage exceeds the threshold value (client-output-buffer-limit-normal-soft-limit) for longer than this parameter value, the client connection is disconnected. <br/> Setting the value too low may cause the MemStore service to fail to provision. <br/> It is recommended to set a sufficient value or the default value (0) considering the workload. |

### min-replicas-max-lag

| Type    | Input range  | Description                                                                                                          |
|---------|--------------|----------------------------------------------------------------------------------------------------------------------|
| integer | 0~2147483647 | Maximum ping response delay time allowed for replica nodes by the primary node. <br/><br/> **Possible issues**: <br/> If the primary node does not receive pings within the set time, the replica is considered unavailable until it reconnects. <br/> It is recommended to set a reasonable value considering service stability. |

### min-replicas-to-write

| Type    | Input range  | Description                                                                                                          |
|---------|--------------|----------------------------------------------------------------------------------------------------------------------|
| integer | 0~2147483647 | The minimum number of replicas the primary node must maintain to accept write requests. <br/><br/> **Possible issues**: <br/> If the primary node does not have the required number of replicas as set by `min-replicas-to-write`, write requests will be rejected on the primary node. <br/> Example: If `min-replicas-to-write: 5` is set but the cluster has only two replica nodes, a write will be rejected: <br/> ```primary.memstore:6379> SET hello memstore(error) NOREPLICAS Not enough good replicas to write.``` |

### repl-backlog-size

| Type    | Input range | Description                                                                                                          |
|---------|-------------|----------------------------------------------------------------------------------------------------------------------|
| integer | 16384~      | Size of the backlog buffer for primary node to synchronize data with replica nodes. <br/><br/> **Possible issues**: <br/> If the replication lag between primary and replica nodes exceeds the backlog size set by this parameter, a full sync occurs. <br/> Setting the value too low may lead to frequent full syncs, which can **affect service performance** or **temporarily reduce service availability**. It is recommended to set a sufficient value considering the workload. |

### repl-backlog-ttl

| Type    | Input range  | Description                                                                                                          |
|---------|--------------|----------------------------------------------------------------------------------------------------------------------|
| integer | 0~2147483647 | Time period for which the primary node retains the backlog buffer for synchronization with replica nodes. <br/> Setting to `0` means the backlog buffer is never released. <br/><br/> **Possible issues**: <br/> After the set time, the backlog data on the primary node is released, and full sync is performed for data replication with the replica node. <br/> Setting the value too low may cause frequent full syncs, which **affects service performance** or temporarily reduces service availability. It is recommended to set a sufficient value considering the workload. |

### maxclients

| Type    | Input range | Description                                                                                                          |
|---------|-------------|----------------------------------------------------------------------------------------------------------------------|
| integer | 100~65000   | Maximum number of clients that can connect to a node. <br/><br/> **Possible issues**: <br/> If the value is set too low for the workload, clients may fail to connect, and the node may fail health checks, causing **failover**. <br/> It is recommended to set a sufficient value considering the node type and workload. |



## Copy parameter group \{#copy-parameter-group}

Custom parameter groups provide a copy feature that can be used for backup. Copy the selected parameter group to create a new parameter group.

1. Go to [KakaoCloud Console](https://console.kakaocloud.com/) > **Data Store > MemStore** menu.
2. Click the **Parameter Group** menu and select the **Custom group** tab.
3. Select the parameter group to copy, then click the [More] icon > **Copy**.
4. In the **Parameter Group Copy** popup, review the input information and click the [OK] button.
5. Verify the copied parameter group in the **Custom group** tab.


## Reset parameter group to default \{#reset-parameter-group-to-default}


Custom parameter groups allow you to reset parameter values back to their initial defaults after modification. The default values refer to the default values provided by the selected engine version when the group was created. Resetting to default will attempt to apply the changes to all connected clusters.

:::info
If the parameter application status of connected clusters is `Pending` or `Applying`, you cannot reset the parameters to default.
:::

1. Go to [KakaoCloud Console](https://console.kakaocloud.com/) > **Data Store > MemStore** menu.
2. Click the **Parameter Group** menu and select the **Custom group** tab.
3. Select the parameter group to reset to default, then click the [More] icon > **Reset parameters to default**.
4. In the **Reset to Default** popup, review the instructions and click the [OK] button. <br/> If multiple clusters are connected to the parameter group, you can select the parameter application method. <br/>- `Apply all`: Apply the modified parameters to all clusters simultaneously. <br/> -`Sequential apply`: Apply the modified parameters to clusters sequentially. The parameter application status for pending clusters will be displayed as `Pending`.
5. If connected clusters are present, check the parameter application status to ensure the reset has been applied successfully to each cluster.


## Roll back parameter group \{#roll-back-parameter-group}

Custom parameter groups provide a rollback feature to revert the parameter values to their previous state after modifications. This feature is only available when there is a modification history just before attempting the rollback.

:::info
- If the parameter application status of connected clusters is `Pending` or `Applying`, the parameter group cannot be rolled back.
- After rollback, you cannot revert to earlier modification histories, as the parameters have already been changed to the most recent state. If you modify the parameters again after rollback, a new modification history will be created, allowing rollback again.
:::

1. Go to [KakaoCloud Console](https://console.kakaocloud.com/) > **Data Store > MemStore** menu.
2. Click the **Parameter Group** menu and select the **Custom group** tab.
3. Select the parameter group to roll back, then click the [More] icon > **Rollback**.
4. In the **Rollback Parameter Group** popup, review the instructions and click the [OK] button.
5. If connected clusters are present, check the parameter application status to ensure the rollback has been applied successfully to each cluster.


## Delete parameter group \{#delete-parameter-group}

You can delete a custom parameter group if it is no longer needed.

:::info
- Default parameter groups cannot be deleted.
- Parameter groups cannot be deleted if they are connected to a cluster.
:::


1. Go to [KakaoCloud Console](https://console.kakaocloud.com/) > **Data Store > MemStore** menu.
2. Click the **Parameter Group** menu and select the **Custom group** tab.
3. Select the parameter group to delete, then click the [More] icon > **Delete**.
4. In the **Delete Parameter Group** popup, enter the required information and click the [OK] button.
5. Verify the parameter group has been deleted in the **Custom group** tab.



## Retry parameter application \{#retry-parameter-application}

If the parameter application status of a connected instance is `Error-Sync` or `Suspended`, you can retry the parameter application.<br/>
For more details on parameter application status, refer to [Parameter application status](/service/data-store/memstore/memstore-main#parameter-application-status).

1. Go to [KakaoCloud Console](https://console.kakaocloud.com/) > **Data Store > MemStore** menu.
2. Click the **Parameter Group** menu and select the **Custom group** tab.
3. Select the parameter group to view detailed information.
3. Go to the **Connected clusters** tab.
4. Click the [Retry] button for the instance you want to retry.