Elasticsearch:Downgrade a cluster

Limits

• You can downgrade a cluster only if the following conditions are met:

  • Downgrade interval

    The interval between two downgrades on the same cluster must be at least 30 minutes.

  • Cluster load

    To ensure cluster stability during a downgrade, the load on different node roles must meet the following conditions.

    Node type	CPU utilization	JVM heap memory usage
    Dedicated master node	Maximum CPU utilization of a single node in the last 24 hours: < 30%	Maximum JVM heap memory usage of a single node in the last 24 hours: < 25%
    Other node roles	Both of the following conditions must be met: Maximum CPU utilization of a single node in the last 24 hours: < 50% Average CPU utilization of all nodes in the last 24 hours: < 30%	Both of the following conditions must be met: Maximum JVM heap memory usage of a single node in the last 24 hours: < 50% Average JVM heap memory usage of all nodes in the last 24 hours: < 30%

  • Specifications

    The target CPU and memory specifications must be greater than or equal to half of the current specifications. You cannot downgrade a node to the following specifications: 1 CPU core and 2 GiB of memory, 2 CPU cores and 2 GiB of memory, 2 CPU cores and 4 GiB of memory, or 4 CPU cores and 4 GiB of memory.

    Note

    • You cannot downgrade Kibana nodes to 1 CPU core and 2 GiB of memory.

    • To downgrade to 2 CPU cores and 4 GiB of memory or 4 CPU cores and 4 GiB of memory, you must create a new instance with the target specifications. Then, migrate the data from the original instance to the new one. For example, you can use Logstash to migrate the data.

  • Disk type downgrade

    You can downgrade disks from a higher performance level to a lower one. The disk types are listed in descending order of performance: enterprise SSDs (ESSDs), standard SSDs, and ultra disks. You can downgrade the disk type as needed. For more information about disks, see Disk overview.

    Note

    You cannot downgrade ESSD PL0 disks to standard SSDs.

• For ES V7.16, V8.5, V8.9, and later, you can reduce the storage space of data nodes. To ensure cluster stability, the disk usage must remain below a safe level during the process. The disk usage after the downgrade must not exceed 60%. This is calculated as follows: (Current disk space used) / (Disk space after downgrade) < 60%.

Precautions

• If your cluster indexes have replica shards and the cluster load is normal, the cluster can typically continue to provide service during the downgrade. A normal load is defined as CPU utilization of approximately 60%, heap memory usage of approximately 50%, and a load_1m value lower than the number of CPU cores.

• Downgrading a cluster triggers a restart. The restart duration depends on factors such as cluster specifications, data structure, and data size. We recommend that you perform the downgrade during off-peak hours. The restart process can be lengthy and may take several hours. For more information, see Restart an instance or node.

• A downgrade does not cause data loss. However, if the cluster load is high or the cluster status is abnormal, you might experience occasional access timeouts or jitter. We recommend implementing a retry mechanism in your application and increasing the number of replicas.

• For clusters with non-standard specifications, such as a cluster with only two data nodes, a downgrade can pose additional risks such as data loss. Proceed with caution.

• Kibana is typically used as a visualization platform for operations such as real-time data queries. Therefore, historical usage checks are not performed when you downgrade Kibana nodes. You can configure and adjust the specifications based on your needs. If usage increases later, you must promptly upgrade to higher specifications.

• After you submit a downgrade order, the instance is billed based on the new configuration. For billing rules, see Pay-as-you-go and Subscription.

  Note

  When you downgrade a cluster, you can view the billing details for the new configuration in real time on the Downgrade page.

Procedure

1. Log on to the Alibaba Cloud Elasticsearch console.

2. In the top menu bar, select a resource group and a region. In the navigation pane on the left, click Elasticsearch Instances.

3. In the instance list, find the target instance. In the Actions column, click the > Downgrade Configuration.

4. Change the node specifications or storage class as needed.

5. Read and agree to the Terms of Service and risk notifications, and then click Buy Now.

   After you complete the payment, the cluster restarts. The downgrade is complete after the cluster successfully restarts.

FAQ

• What do I do if I cannot upgrade or downgrade my cluster?

• Can I change the disk type for an ES instance?

• Does changing the cluster configuration affect the ES service?

• After the number of nodes is changed, does the cluster automatically rebalance shards?

• What do I do if I selected the wrong configuration when purchasing an ES instance?

• After I upgrade an instance, can I downgrade it? If so, how?

• Why can't I downgrade cold data nodes?

References

• For information about the API operation used to downgrade a cluster, see UpdateInstance.

• See Specification and capacity evaluation and Limits to ensure that your cluster has sufficient storage space after the downgrade. If the cluster load is high after the downgrade, you must promptly upgrade the cluster. For more information, see Upgrade a cluster.