Restrictions and known problemsedit

When using Elasticsearch Service, there are some limitations you should be aware of:

For limitations related to logging and monitoring, check the Restrictions and limitations section of the logging and monitoring page.

Occasionally, we also publish information about Known problems with our Elasticsearch Service or the Elastic Stack.

To learn more about the features that are supported by Elasticsearch Service, check Elastic Cloud Subscriptions.


  • File and LDAP realms cannot be used. The Native realm is enabled, but the realm configuration itself is fixed in Elastic Cloud. Alternatively, authentication protocols such as SAML, OpenID Connect, or Kerberos can be used.
  • Client certificates, such as PKI certificates, are not supported.


The following restrictions apply when using APIs in Elasticsearch Service:

Elasticsearch Service API
The Elasticsearch Service API is subject to a restriction on the volume of API requests that can be submitted per user, per second. Check Rate limiting for details.
Elasticsearch APIs
The Elasticsearch APIs do not natively enforce rate limiting. However, all requests to the Elasticsearch cluster are subject to Elasticsearch configuration settings, such as the network HTTP setting http:max_content_length which restricts the maximum size of an HTTP request body. This setting has a default value of 100MB, hence restricting API request payloads to that size. This setting is not currently configurable in Elasticsearch Service. For a list of which Elasticsearch settings are supported on Cloud, check Add Elasticsearch user settings. To learn about using the Elasticsearch APIs in Elasticsearch Service, check Access the Elasticsearch API console. And, for full details about the Elasticsearch APIs and their endpoints, check the Elasticsearch API reference documentation.
Kibana APIs
There are no rate limits restricting your use of the Kibana APIs. However, Kibana features are affected by the Kibana configuration settings, not all of which are supported in Elasticsearch Service. For a list of what settings are currently supported, check Add Kibana user settings. For all details about using the Kibana APIs, check the Kibana API reference documentation.

Transport clientedit

  • The transport client is not considered thread safe in a cloud environment. We recommend that you use the Java REST client instead. This restriction relates to the fact that your deployments hosted on Elasticsearch Service are behind proxies, which prevent the transport client from communicating directly with Elasticsearch clusters.
  • The transport client is not supported over private link connections. Use the Java REST client instead, or connect over the public internet.
  • The transport client does not work with Elasticsearch clusters at version 7.6 and later that are hosted on Cloud. Transport client continues to work with Elasticsearch clusters at version 7.5 and earlier. Note that the transport client was deprecated with version 7.0 and will be removed with 8.0.

Elasticsearch and Kibana pluginsedit

  • Kibana plugins are not supported.
  • Elasticsearch plugins, are not enabled by default for security purposes. Please reach out to support if you would like to enable Elasticsearch plugins support on your account.
  • Some Elasticsearch plugins do not apply to Elasticsearch Service. For example, you won’t ever need to change discovery, as Elasticsearch Service handles how nodes discover one another.
  • In Elasticsearch 5.0 and later, site plugins are no longer supported. This change does not affect the site plugins Elasticsearch Service might provide out of the box, such as Kopf or Head, since these site plugins are serviced by our proxies and not Elasticsearch itself.
  • In Elasticsearch 5.0 and later, site plugins such as Kopf and Paramedic are no longer provided. We recommend that you use our cluster performance metrics, X-Pack monitoring features and Kibana’s (6.3+) Index Management UI if you want more detailed information or perform index management actions.


Watcher encryption Key Setup is not supported.

Changing the default throttle period is not possible. You can specify a throttle period per watch, however.

Watcher comes preconfigured with a directly usable email account provided by Elastic. However, this account can’t be reconfigured and is subject to some limitations. For more information on the limits of the Elastic mail server, check the cloud email service limits

Alternatively, a custom mail server can be configured as described in Configuring a custom mail server

Private Link and SSO to Kibana URLsedit

Currently you can’t use SSO to login directly from Elastic Cloud into Kibana endpoints that are protected by Private Link traffic filters. However, you can still SSO into Private Link protected Kibana endpoints individually using the SAML or OIDC protocol from your own identity provider, just not through the Elastic Cloud console. Stack level authentication using the Elasticsearch username and password should also work with {kibana-id}.{vpce|privatelink|psc}.domain URLs.

Traffic filters and Watcher Webhooks / PDF report generationedit

You can’t set the Watcher webhook target to a deployment protected by traffic filters. This restriction also applies to generating PDFs using Watcher. Public webhooks such as Slack, PagerDuty, and email are not affected by this limitation. You can use the xpack.notification.webhook.additional_token_enabled configuration setting available in Elastic Stack versions 8.7.1 and later to bypass traffic filters.


  • The maximum size of a single Kibana instance is 8GB. This means, Kibana instances can be scaled up to 8GB before they are scaled out. For example, when creating a deployment with a Kibana instance of size 16GB, then 2x8GB instances are created. If you face performance issues with Kibana PNG or PDF reports, the recommendations are to create multiple, smaller dashboards to export the data, or to use a third party browser extension for exporting the dashboard in the format you need.
  • Running an external Kibana in parallel to Elasticsearch Service’s Kibana instances may cause errors, for example Unable to decrypt attribute, due to a mismatched xpack.encryptedSavedObjects.encryptionKey as Elasticsearch Service does not allow users to set nor expose this value. While workarounds are possible, this is not officially supported nor generally recommended.

APM Agent central configuration with PrivateLink or traffic filtersedit

If you are using APM 7.9.0 or older:

Fleet with PrivateLink or traffic filtersedit

If you are using Fleet 7.13.x:

  • You cannot use Fleet 7.13.x, if your deployment is secured by traffic filters. Fleet 7.14.0 and later works with traffic filters (both Private Link and IP filters).

Enterprise Search in Kibanaedit

Enterprise Search’s management interface in Kibana does not work with traffic filters with 8.3.1 and older, it will return an Insufficient permissions (403 Forbidden) error. In Kibana 8.3.2, 8.4.0 and higher, the Enterprise Search management interface works with traffic filters.

Restoring a snapshot across deploymentsedit

Kibana and Enterprise Search do not currently support restoring a snapshot of their indices across Elastic Cloud deployments.

Kibana uses encryption keys in various places, ranging from encrypting data in some areas of reporting, alerts, actions, connector tokens, ingest outputs used in Fleet and Synthetics monitoring to user sessions.

Enterprise Search uses encryption keys when storing content source synchronization credentials, API tokens and other sensitive information.

Currently, there is not a way to retrieve the values of Kibana and Enterprise Search encryption keys, or set them in the target deployment before restoring a snapshot. As a result, once a snapshot is restored, Kibana and Enterprise Search will not be able to decrypt the data required for some Kibana and Enterprise Search features to function properly in the target deployment.

If you have already restored a snapshot across deployments and now have broken Kibana saved objects or Enterprise Search features in the target deployment, you will have to recreate all broken configurations and objects, or create a new setup in the target deployment instead of using snapshot restore.

Migrate Fleet-managed Elastic Agents across deployments by restoring a snapshotedit

There are situations where you may need or want to move your installed Elastic Agents from being managed in one deployment to being managed in another deployment.

In Elastic Cloud, you can migrate your Elastic Agents by taking a snapshot of your source deployment, and restoring it on a target deployment.

To make a seamless migration, after restoring from a snapshot there are some additional steps required, such as updating settings and resetting the agent policy. Check Migrate Elastic Agents for details.

Regions and Availability Zonesedit

  • The AWS us-west-1 region is limited to two availability zones for ES data nodes and one (tiebreaker only) virtual zone (as depicted by the -z in the AZ (us-west-1z). Deployment creation with three availability zones for Elasticsearch data nodes for hot, warm, and cold tiers is not possible. This includes scaling an existing deployment with one or two AZs to three availability zones. The virtual zone us-west-1z can only hold an Elasticsearch tiebreaker node (no data nodes). The workaround is to use a different AWS US region that allows three availability zones, or to scale existing nodes up within the two availability zones.
  • The AWS eu-central-2 region is limited to two availability zones for CPU Optimized (ARM) Hardware profile ES data node and warm/cold tier. Deployment creation with three availability zones for Elasticsearch data nodes for hot (for CPU Optimized (ARM) profile), warm and cold tiers is not possible. This includes scaling an existing deployment with one or two AZs to three availability zones. The workaround is to use a different AWS region that allows three availability zones, or to scale existing nodes up within the two availability zones.

Known problemsedit

  • There is a known problem affecting clusters with versions 7.7.0 and 7.7.1 due to a bug in Elasticsearch. Although rare, this bug can prevent you from running plans. If this occurs we recommend that you retry the plan, and if that fails please contact support to get your plan through. Because of this bug we recommend you to upgrade to version 7.8 and higher, where the problem has already been addressed.
  • A known issue can prevent direct rolling upgrades from Elasticsearch version 5.6.10 to version 6.3.0. As a workaround, we have removed version 6.3.0 from the Elasticsearch Service Console for new cluster deployments and for upgrading existing ones. If you are affected by this issue, check Rolling upgrades from 5.6.x to 6.3.0 fails with "java.lang.IllegalStateException: commit doesn’t contain history uuid" in our Elastic Support Portal. If these steps do not work or you do not have access to the Support Portal, you can contact

Repository Analysis API is unavailable in Elastic Cloudedit

  • The Elasticsearch Repository analysis API is not available in Elastic Cloud due to deployments defaulting to having operator privileges enabled that prevent non-operator privileged users from using it along with a number of other APIs.