Updating custom templates to support node_roles and autoscaling
editUpdating custom templates to support node_roles
and autoscaling
editCustom deployment templates should be updated in order to take advantage of new Elastic Cloud Enterprise features, such as Data tiers (that is, the new cold and frozen data tiers) and Deployment autoscaling.
By updating these templates we also ensure forward compatibility with future Elastic Cloud Enterprise versions that will require certain fields such as node_roles
and id
to be present in the deployment configuration.
System owned deployment templates have already been updated to support both data tiers with node_roles
and autoscaling.
However, the custom templates that you created need to be manually updated by following the steps in this guide.
Adding support for node_roles
editThe node_roles
field defines the roles that an Elasticsearch topology element can have, which is used in place of node_type
when a new feature such as autoscaling is enabled, or when a new data tier is added.
This field is supported on Elastic stack versions 7.10 and above.
There are a number of fields that need to be added to each Elasticsearch node in order to support node_roles
:
-
id: Unique identifier of the topology element. This field, along with the
node_roles
, identifies an Elasticsearch topology element. -
node_roles: The list of node roles. Allowable roles are:
master
,ingest
,ml
,data_hot
,data_content
,data_warm
,data_cold
,data_frozen
,remote_cluster_client
, andtransform
. For details, check Node roles. -
topology_element_control: Controls for the topology element.
-
min: The absolute minimum size limit for a topology element.
If the value is
0
, that means the topology element can be disabled.
-
min: The absolute minimum size limit for a topology element.
If the value is
The following example is based on the default
system owned deployment template that already supports node_roles
. This template will be used as a reference for the next sections:
Reference example with support for node_roles
{ ... "deployment_template": { "resources": { "elasticsearch": [ { "plan": { "cluster_topology": [ { "id": "hot_content", "instance_configuration_id": "data.default", "zone_count": 1, "node_roles": [ "master", "ingest", "data_hot", "data_content", "remote_cluster_client", "transform" ], "node_type": { "master": true, "data": true, "ingest": true }, "elasticsearch": { "node_attributes": { "data": "hot" } }, "size": { "value": 4096, "resource": "memory" }, "topology_element_control": { "min": { "value": 1024, "resource": "memory" } } }, { "id": "warm", "instance_configuration_id": "data.highstorage", "zone_count": 1, "node_roles": [ "data_warm", "remote_cluster_client" ], "node_type": { "data": true, "ingest": false, "master": false }, "elasticsearch": { "node_attributes": { "data": "warm" } }, "size": { "resource": "memory", "value": 0 }, "topology_element_control": { "min": { "value": 0, "resource": "memory" } } }, { "id": "cold", "instance_configuration_id": "data.highstorage", "zone_count": 1, "node_roles": [ "data_cold", "remote_cluster_client" ], "node_type": { "data": true, "ingest": false, "master": false }, "elasticsearch": { "node_attributes": { "data": "cold" } }, "size": { "resource": "memory", "value": 0 }, "topology_element_control": { "min": { "value": 0, "resource": "memory" } } }, { "id": "frozen", "instance_configuration_id": "data.frozen", "zone_count": 1, "node_roles": [ "data_frozen" ], "node_type": { "data": true, "ingest": false, "master": false }, "elasticsearch": { "node_attributes": { "data": "frozen" } }, "size": { "resource": "memory", "value": 0 }, "topology_element_control": { "min": { "value": 0, "resource": "memory" } } }, { "id": "coordinating", "zone_count": 1, "instance_configuration_id": "coordinating", "node_roles": [ "ingest", "remote_cluster_client" ], "node_type": { "master": false, "data": false, "ingest": true }, "size": { "value": 0, "resource": "memory" }, "topology_element_control": { "min": { "value": 0, "resource": "memory" } } }, { "id": "master", "zone_count": 1, "instance_configuration_id": "master", "node_roles": [ "master", "remote_cluster_client" ], "node_type": { "master": true, "data": false, "ingest": false }, "size": { "value": 0, "resource": "memory" }, "topology_element_control": { "min": { "value": 0, "resource": "memory" } } }, { "id": "ml", "zone_count": 1, "instance_configuration_id": "ml", "node_roles": [ "ml", "remote_cluster_client" ], "node_type": { "master": false, "data": false, "ingest": false, "ml": true }, "size": { "value": 0, "resource": "memory" }, "topology_element_control": { "min": { "value": 0, "resource": "memory" } } } ], "elasticsearch": {} }, ... } ], "kibana": [ { "ref_id": "kibana-ref-id", "elasticsearch_cluster_ref_id": "es-ref-id", "region": "ece-region", "plan": { "zone_count": 1, "cluster_topology": [ { "instance_configuration_id": "kibana", "size": { "value": 1024, "resource": "memory" } } ], "kibana": {} } } ], "apm": [ { "ref_id": "apm-ref-id", "elasticsearch_cluster_ref_id": "es-ref-id", "region": "ece-region", "plan": { "cluster_topology": [ { "instance_configuration_id": "apm", "size": { "value": 0, "resource": "memory" }, "zone_count": 1 } ], "apm": {} } } ], "enterprise_search": [ { "ref_id": "enterprise_search-ref-id", "elasticsearch_cluster_ref_id": "es-ref-id", "region": "ece-region", "plan": { "cluster_topology": [ { "node_type": { "appserver": true, "connector": true, "worker": true }, "instance_configuration_id": "enterprise.search", "size": { "value": 0, "resource": "memory" }, "zone_count": 2 } ], "enterprise_search": {} } } ] } } }
In the reference example there are seven different Elasticsearch topology elements: hot_content
, warm
, cold
, frozen
, coordinating
, master
, and ml
.
These names map to the id
field of each topology element.
In addition, this template contains four different resources: elasticsearch
, kibana
, apm
, and enterprise_search
.
Requirements
editTo add support for node_roles
, the deployment template has to meet the following requirements:
-
Contains all four
resources
:elasticsearch
,kibana
,apm
, andenterprise_search
. -
The
elasticsearch
resource contains all seven topology elements:hot_content
,warm
,cold
,frozen
,coordinating
,master
, andml
.The IDs
hot_content
,warm
,cold
,frozen
,coordinating
,master
, andml
are the only ones supported in an Elasticsearch topology element. In addition, there may not be topology elements with duplicate IDs inside theelasticsearch
resource. -
Each topology element contains the
id
,node_roles
, andtopology_element_control
fields.
It is also recommended that all Elasticsearch topology elements have the node_attributes
field. This field can be useful in ILM
policies, when creating a deployment using a version below 7.10, that does not support node_roles
.
Except for the id
and node_roles
, all fields can be configured by the user. Also, the topology elements must contain the exact same id
and node_roles
that are present in the reference example.
Although it is required for the template to contain all resources and topology elements, it is possible to disable certain
components by setting their size.value
(and topology_element_control.size
in the case of the topology elements) to 0
.
Updating an ECE custom template to support node_roles
editTo update a custom deployment template:
-
Add the
id
,node_roles
,node_attributes
, andtopology_element_control
fields to the existing Elasticsearch topology elements. Keep in mind that these fields must match the Elasticsearch topology element in question:-
If the
id
of the topology elements in the existing templates already match any of the seven mentioned in the requirements, then simply add thenode_roles
andtopology_element_control
to those elements, based on the reference example. -
Otherwise, map each of the existing topology elements to one of the seven Elasticsearch topology elements, based on their
node_type
, and add the fields accordingly.
-
If the
-
Add the
elasticsearch
topology elements that are missing. -
Add the
resources
that are missing.
Example
editThe existing template contains three Elasticsearch topology elements and two resources (elasticsearch
and kibana
).
Custom example without support for node_roles
{ ... "deployment_template": { "resources": { "elasticsearch": [ { "plan": { "cluster_topology": [ { "instance_configuration_id": "custom.data", "zone_count": 2, "node_type": { "master": true, "data": true, "ingest": true }, "size": { "value": 8192, "resource": "memory" }, "topology_element_control": { "min": { "value": 1024, "resource": "memory" } } }, { "zone_count": 1, "instance_configuration_id": "custom.master", "node_type": { "master": true, "data": false, "ingest": false }, "size": { "value": 0, "resource": "memory" }, "topology_element_control": { "min": { "value": 0, "resource": "memory" } } }, { "zone_count": 1, "instance_configuration_id": "custom.ml", "node_type": { "master": false, "data": false, "ingest": false, "ml": true }, "size": { "value": 0, "resource": "memory" }, "topology_element_control": { "min": { "value": 0, "resource": "memory" } } } ], "elasticsearch": {} }, ... } ], "kibana": [ { "ref_id": "kibana-ref-id", "elasticsearch_cluster_ref_id": "es-ref-id", "region": "ece-region", "plan": { "zone_count": 1, "cluster_topology": [ { "instance_configuration_id": "kibana", "size": { "value": 1024, "resource": "memory" } } ], "kibana": {} } } ] } } }
In this case we can match the three existing Elasticsearch topology elements to hot_content
, master
, and ml
, respectively, based on their node_type
field. Therefore,
we can simply add the id
, node_roles
, topology_element_control
, and node_attributes
that are already associated
with these topology elements in the reference example.
Then, it is only necessary to add the four Elasticsearch topology elements (warm
, cold
, frozen
, and coordinating
) and two resources
(apm
and enterprise_search
) that are missing. These fields can also be added based on the reference example.
After adding support for node_roles
, the resulting deployment template should look similar to the following:
Custom example with support for node_roles
{ ... "deployment_template": { "resources": { "elasticsearch": [ { "plan": { "cluster_topology": [ { "id": "hot_content", "instance_configuration_id": "custom.data", "zone_count": 2, "node_roles": [ "master", "ingest", "data_hot", "data_content", "remote_cluster_client", "transform" ], "node_type": { "master": true, "data": true, "ingest": true }, "elasticsearch": { "node_attributes": { "data": "hot" } }, "size": { "value": 8192, "resource": "memory" }, "topology_element_control": { "min": { "value": 1024, "resource": "memory" } } }, { "id": "warm", "instance_configuration_id": "data.highstorage", "zone_count": 1, "node_roles": [ "data_warm", "remote_cluster_client" ], "node_type": { "data": true, "ingest": false, "master": false }, "elasticsearch": { "node_attributes": { "data": "warm" } }, "size": { "resource": "memory", "value": 0 }, "topology_element_control": { "min": { "value": 0, "resource": "memory" } } }, { "id": "cold", "instance_configuration_id": "data.highstorage", "zone_count": 1, "node_roles": [ "data_cold", "remote_cluster_client" ], "node_type": { "data": true, "ingest": false, "master": false }, "elasticsearch": { "node_attributes": { "data": "cold" } }, "size": { "resource": "memory", "value": 0 }, "topology_element_control": { "min": { "value": 0, "resource": "memory" } } }, { "id": "frozen", "instance_configuration_id": "data.frozen", "zone_count": 1, "node_roles": [ "data_frozen" ], "node_type": { "data": true, "ingest": false, "master": false }, "elasticsearch": { "node_attributes": { "data": "frozen" } }, "size": { "resource": "memory", "value": 0 }, "topology_element_control": { "min": { "value": 0, "resource": "memory" } } }, { "id": "coordinating", "zone_count": 1, "instance_configuration_id": "coordinating", "node_roles": [ "ingest", "remote_cluster_client" ], "node_type": { "master": false, "data": false, "ingest": true }, "size": { "value": 0, "resource": "memory" }, "topology_element_control": { "min": { "value": 0, "resource": "memory" } } }, { "id": "master", "zone_count": 1, "instance_configuration_id": "custom.master", "node_roles": [ "master", "remote_cluster_client" ], "node_type": { "master": true, "data": false, "ingest": false }, "size": { "value": 0, "resource": "memory" }, "topology_element_control": { "min": { "value": 0, "resource": "memory" } } }, { "id": "ml", "zone_count": 1, "instance_configuration_id": "custom.ml", "node_roles": [ "ml", "remote_cluster_client" ], "node_type": { "master": false, "data": false, "ingest": false, "ml": true }, "size": { "value": 0, "resource": "memory" }, "topology_element_control": { "min": { "value": 0, "resource": "memory" } } } ], "elasticsearch": {} }, ... } ], "kibana": [ { "ref_id": "kibana-ref-id", "elasticsearch_cluster_ref_id": "es-ref-id", "region": "ece-region", "plan": { "zone_count": 1, "cluster_topology": [ { "instance_configuration_id": "kibana", "size": { "value": 1024, "resource": "memory" } } ], "kibana": {} } } ], "apm": [ { "ref_id": "apm-ref-id", "elasticsearch_cluster_ref_id": "es-ref-id", "region": "ece-region", "plan": { "cluster_topology": [ { "instance_configuration_id": "apm", "size": { "value": 0, "resource": "memory" }, "zone_count": 1 } ], "apm": {} } } ], "enterprise_search": [ { "ref_id": "enterprise_search-ref-id", "elasticsearch_cluster_ref_id": "es-ref-id", "region": "ece-region", "plan": { "cluster_topology": [ { "node_type": { "appserver": true, "connector": true, "worker": true }, "instance_configuration_id": "enterprise.search", "size": { "value": 0, "resource": "memory" }, "zone_count": 2 } ], "enterprise_search": {} } } ] } } }
Adding support for autoscaling
editAfter adding support for node_roles
we can then update the template to support autoscaling.
Autoscaling is used to automatically adjust the available resources in the deployments.
Currently, this feature is available for Elasticsearch data tiers and machine learning node in Elastic stack versions 7.11 and above.
There are a number of autoscaling fields that need to be added in order to support autoscaling:
- autoscaling_min: The default minimum size of an Elasticsearch topology element when autoscaling is enabled. This setting is currently available only for machine learning nodes, since these are the only nodes that support scaling down.
- autoscaling_max: The default maximum size of an Elasticsearch topology element when autoscaling is enabled. This setting is currently available only for data tiers and machine learning nodes, since these are the only nodes that support scaling up.
-
autoscaling_enabled: When set to
true
, autoscaling is enabled by default on an Elasticsearch cluster.
These fields represent the default settings for the deployment. However, autoscaling can be enabled/disabled and the maximum and minimum autoscaling sizes can be adjusted in the deployment settings.
Similar to the node_roles
example, the following one is also based on the default
deployment template that already supports node_roles
and autoscaling.
This template will be used as a reference for the next sections:
Reference example with support for node_roles
and autoscaling
{ ... "deployment_template": { "resources": { "elasticsearch": [ { "plan": { "cluster_topology": [ { "id": "hot_content", "instance_configuration_id": "data.default", "zone_count": 1, "node_roles": [ "master", "ingest", "data_hot", "data_content", "remote_cluster_client", "transform" ], "node_type": { "master": true, "data": true, "ingest": true }, "elasticsearch": { "node_attributes": { "data": "hot" } }, "size": { "value": 4096, "resource": "memory" }, "topology_element_control": { "min": { "value": 1024, "resource": "memory" } }, "autoscaling_max": { "value": 2097152, "resource": "memory" } }, { "id": "warm", "instance_configuration_id": "data.highstorage", "zone_count": 1, "node_roles": [ "data_warm", "remote_cluster_client" ], "node_type": { "data": true, "ingest": false, "master": false }, "elasticsearch": { "node_attributes": { "data": "warm" } }, "size": { "resource": "memory", "value": 0 }, "topology_element_control": { "min": { "value": 0, "resource": "memory" } }, "autoscaling_max": { "value": 2097152, "resource": "memory" } }, { "id": "cold", "instance_configuration_id": "data.highstorage", "zone_count": 1, "node_roles": [ "data_cold", "remote_cluster_client" ], "node_type": { "data": true, "ingest": false, "master": false }, "elasticsearch": { "node_attributes": { "data": "cold" } }, "size": { "resource": "memory", "value": 0 }, "topology_element_control": { "min": { "value": 0, "resource": "memory" } }, "autoscaling_max": { "value": 2097152, "resource": "memory" } }, { "id": "frozen", "instance_configuration_id": "data.frozen", "zone_count": 1, "node_roles": [ "data_frozen" ], "node_type": { "data": true, "ingest": false, "master": false }, "elasticsearch": { "node_attributes": { "data": "frozen" } }, "size": { "resource": "memory", "value": 0 }, "topology_element_control": { "min": { "value": 0, "resource": "memory" } }, "autoscaling_max": { "value": 2097152, "resource": "memory" } }, { "id": "coordinating", "zone_count": 1, "instance_configuration_id": "coordinating", "node_roles": [ "ingest", "remote_cluster_client" ], "node_type": { "master": false, "data": false, "ingest": true }, "size": { "value": 0, "resource": "memory" }, "topology_element_control": { "min": { "value": 0, "resource": "memory" } } }, { "id": "master", "zone_count": 1, "instance_configuration_id": "master", "node_roles": [ "master", "remote_cluster_client" ], "node_type": { "master": true, "data": false, "ingest": false }, "size": { "value": 0, "resource": "memory" }, "topology_element_control": { "min": { "value": 0, "resource": "memory" } } }, { "id": "ml", "zone_count": 1, "instance_configuration_id": "ml", "node_roles": [ "ml", "remote_cluster_client" ], "node_type": { "master": false, "data": false, "ingest": false, "ml": true }, "size": { "value": 0, "resource": "memory" }, "topology_element_control": { "min": { "value": 0, "resource": "memory" } }, "autoscaling_min": { "resource": "memory", "value": 0 }, "autoscaling_max": { "value": 2097152, "resource": "memory" } } ], "elasticsearch": {}, "autoscaling_enabled": true }, ... } ], "kibana": [ { "ref_id": "kibana-ref-id", "elasticsearch_cluster_ref_id": "es-ref-id", "region": "ece-region", "plan": { "zone_count": 1, "cluster_topology": [ { "instance_configuration_id": "kibana", "size": { "value": 1024, "resource": "memory" } } ], "kibana": {} } } ], "apm": [ { "ref_id": "apm-ref-id", "elasticsearch_cluster_ref_id": "es-ref-id", "region": "ece-region", "plan": { "cluster_topology": [ { "instance_configuration_id": "apm", "size": { "value": 0, "resource": "memory" }, "zone_count": 1 } ], "apm": {} } } ], "enterprise_search": [ { "ref_id": "enterprise_search-ref-id", "elasticsearch_cluster_ref_id": "es-ref-id", "region": "ece-region", "plan": { "cluster_topology": [ { "node_type": { "appserver": true, "connector": true, "worker": true }, "instance_configuration_id": "enterprise.search", "size": { "value": 0, "resource": "memory" }, "zone_count": 2 } ], "enterprise_search": {} } } ] } } }
Requirements
editTo add support for autoscaling, the deployment template has to meet the following requirements:
-
Already has support for
node_roles
. -
Contains the
size
,autoscaling_min
, andautoscaling_max
fields, according to the rules specified in the autoscaling requirements table. -
Contains the
autoscaling_enabled
fields on theelasticsearch
resource.
If necessary, the values chosen for each field can be based on the reference example.
Updating an ECE custom template to support autoscaling
editTo update a custom deployment template:
-
Add the
autoscaling_min
andautoscaling_max
fields to the Elasticsearch topology elements (check Autoscaling through the API). -
Add the
autoscaling_enabled
fields to theelasticsearch
resource. Set this field totrue
in case you want autoscaling enabled by default, and tofalse
otherwise.
Example
editAfter adding support for autoscaling to the example presented in the previous section, the resulting deployment template should look similar to the following:
Custom example with support for node_roles
and autoscaling
{ ... "deployment_template": { "resources": { "elasticsearch": [ { "plan": { "cluster_topology": [ { "id": "hot_content", "instance_configuration_id": "custom.data", "zone_count": 2, "node_roles": [ "master", "ingest", "data_hot", "data_content", "remote_cluster_client", "transform" ], "node_type": { "master": true, "data": true, "ingest": true }, "elasticsearch": { "node_attributes": { "data": "hot" } }, "size": { "value": 8192, "resource": "memory" }, "topology_element_control": { "min": { "value": 1024, "resource": "memory" } }, "autoscaling_max": { "value": 2097152, "resource": "memory" } }, { "id": "warm", "instance_configuration_id": "data.highstorage", "zone_count": 1, "node_roles": [ "data_warm", "remote_cluster_client" ], "node_type": { "data": true, "ingest": false, "master": false }, "elasticsearch": { "node_attributes": { "data": "warm" } }, "size": { "resource": "memory", "value": 0 }, "topology_element_control": { "min": { "value": 0, "resource": "memory" } }, "autoscaling_max": { "value": 2097152, "resource": "memory" } }, { "id": "cold", "instance_configuration_id": "data.highstorage", "zone_count": 1, "node_roles": [ "data_cold", "remote_cluster_client" ], "node_type": { "data": true, "ingest": false, "master": false }, "elasticsearch": { "node_attributes": { "data": "cold" } }, "size": { "resource": "memory", "value": 0 }, "topology_element_control": { "min": { "value": 0, "resource": "memory" } }, "autoscaling_max": { "value": 2097152, "resource": "memory" } }, { "id": "frozen", "instance_configuration_id": "data.frozen", "zone_count": 1, "node_roles": [ "data_frozen" ], "node_type": { "data": true, "ingest": false, "master": false }, "elasticsearch": { "node_attributes": { "data": "frozen" } }, "size": { "resource": "memory", "value": 0 }, "topology_element_control": { "min": { "value": 0, "resource": "memory" } }, "autoscaling_max": { "value": 2097152, "resource": "memory" } }, { "id": "coordinating", "zone_count": 1, "instance_configuration_id": "coordinating", "node_roles": [ "ingest", "remote_cluster_client" ], "node_type": { "master": false, "data": false, "ingest": true }, "size": { "value": 0, "resource": "memory" }, "topology_element_control": { "min": { "value": 0, "resource": "memory" } } }, { "id": "master", "zone_count": 1, "instance_configuration_id": "custom.master", "node_roles": [ "master", "remote_cluster_client" ], "node_type": { "master": true, "data": false, "ingest": false }, "size": { "value": 0, "resource": "memory" }, "topology_element_control": { "min": { "value": 0, "resource": "memory" } } }, { "id": "ml", "zone_count": 1, "instance_configuration_id": "custom.ml", "node_roles": [ "ml", "remote_cluster_client" ], "node_type": { "master": false, "data": false, "ingest": false, "ml": true }, "size": { "value": 0, "resource": "memory" }, "topology_element_control": { "min": { "value": 0, "resource": "memory" } }, "autoscaling_min": { "resource": "memory", "value": 0 }, "autoscaling_max": { "value": 2097152, "resource": "memory" } } ], "elasticsearch": {}, "autoscaling_enabled": true }, ... } ], "kibana": [ { "ref_id": "kibana-ref-id", "elasticsearch_cluster_ref_id": "es-ref-id", "region": "ece-region", "plan": { "zone_count": 1, "cluster_topology": [ { "instance_configuration_id": "kibana", "size": { "value": 1024, "resource": "memory" } } ], "kibana": {} } } ], "apm": [ { "ref_id": "apm-ref-id", "elasticsearch_cluster_ref_id": "es-ref-id", "region": "ece-region", "plan": { "cluster_topology": [ { "instance_configuration_id": "apm", "size": { "value": 0, "resource": "memory" }, "zone_count": 1 } ], "apm": {} } } ], "enterprise_search": [ { "ref_id": "enterprise_search-ref-id", "elasticsearch_cluster_ref_id": "es-ref-id", "region": "ece-region", "plan": { "cluster_topology": [ { "node_type": { "appserver": true, "connector": true, "worker": true }, "instance_configuration_id": "enterprise.search", "size": { "value": 0, "resource": "memory" }, "zone_count": 2 } ], "enterprise_search": {} } } ] } } }
Updating a custom template through the RESTful API
editHaving added support for node_roles
and autoscaling to your custom template, it is possible to perform the update through the RESTful API, by following these steps:
-
Obtain the existing deployment templates by sending the following
GET
request, and take note of theid
of the template you wish to update.curl -k -X GET -H "Authorization: ApiKey $ECE_API_KEY" https://COORDINATOR_HOST:12443/api/v1/deployments/templates?region=ece-region
-
Send a
PUT
request with the updated template on the payload, in order to effectively replace the outdated template with the new one. Note that the following request is just an example, you have to replace{template_id}
with theid
you collected on step 1. and set the payload to the updated template JSON. Check set deployment template API for more details.Update template API request example
curl -k -X PUT -H "Authorization: ApiKey $ECE_API_KEY" https://$COORDINATOR_HOST:12443/api/v1/deployments/templates/{template_id}?region=ece-region -H 'content-type: application/json' -d ' { "name": "ECE Custom Template", "description": "ECE custom template with support for node_roles and autoscaling", "deployment_template": { "resources": { "elasticsearch": [ { "ref_id": "es-ref-id", "region": "ece-region", "plan": { "cluster_topology": [ { "id": "hot_content", "instance_configuration_id": "custom.data", "zone_count": 2, "node_roles": [ "master", "ingest", "data_hot", "data_content", "remote_cluster_client", "transform" ], "node_type": { "master": true, "data": true, "ingest": true }, "elasticsearch": { "node_attributes": { "data": "hot" } }, "size": { "value": 8192, "resource": "memory" }, "topology_element_control": { "min": { "value": 1024, "resource": "memory" } }, "autoscaling_max": { "value": 2097152, "resource": "memory" } }, { "id": "warm", "instance_configuration_id": "data.highstorage", "zone_count": 1, "node_roles": [ "data_warm", "remote_cluster_client" ], "node_type": { "data": true, "ingest": false, "master": false }, "elasticsearch": { "node_attributes": { "data": "warm" } }, "size": { "resource": "memory", "value": 0 }, "topology_element_control": { "min": { "value": 0, "resource": "memory" } }, "autoscaling_max": { "value": 2097152, "resource": "memory" } }, { "id": "cold", "instance_configuration_id": "data.highstorage", "zone_count": 1, "node_roles": [ "data_cold", "remote_cluster_client" ], "node_type": { "data": true, "ingest": false, "master": false }, "elasticsearch": { "node_attributes": { "data": "cold" } }, "size": { "resource": "memory", "value": 0 }, "topology_element_control": { "min": { "value": 0, "resource": "memory" } }, "autoscaling_max": { "value": 2097152, "resource": "memory" } }, { "id": "frozen", "instance_configuration_id": "data.frozen", "zone_count": 1, "node_roles": [ "data_frozen" ], "node_type": { "data": true, "ingest": false, "master": false }, "elasticsearch": { "node_attributes": { "data": "frozen" } }, "size": { "resource": "memory", "value": 0 }, "topology_element_control": { "min": { "value": 0, "resource": "memory" } }, "autoscaling_max": { "value": 2097152, "resource": "memory" } }, { "id": "coordinating", "zone_count": 1, "instance_configuration_id": "coordinating", "node_roles": [ "ingest", "remote_cluster_client" ], "node_type": { "master": false, "data": false, "ingest": true }, "size": { "value": 0, "resource": "memory" }, "topology_element_control": { "min": { "value": 0, "resource": "memory" } } }, { "id": "master", "zone_count": 1, "instance_configuration_id": "custom.master", "node_roles": [ "master", "remote_cluster_client" ], "node_type": { "master": true, "data": false, "ingest": false }, "size": { "value": 0, "resource": "memory" }, "topology_element_control": { "min": { "value": 0, "resource": "memory" } } }, { "id": "ml", "zone_count": 1, "instance_configuration_id": "custom.ml", "node_roles": [ "ml", "remote_cluster_client" ], "node_type": { "master": false, "data": false, "ingest": false, "ml": true }, "size": { "value": 0, "resource": "memory" }, "topology_element_control": { "min": { "value": 0, "resource": "memory" } }, "autoscaling_min": { "resource": "memory", "value": 0 }, "autoscaling_max": { "value": 2097152, "resource": "memory" } } ], "elasticsearch": {}, "autoscaling_enabled": true }, "settings": { "dedicated_masters_threshold": 6 } } ], "kibana": [ { "ref_id": "kibana-ref-id", "elasticsearch_cluster_ref_id": "es-ref-id", "region": "ece-region", "plan": { "zone_count": 1, "cluster_topology": [ { "instance_configuration_id": "kibana", "size": { "value": 1024, "resource": "memory" } } ], "kibana": {} } } ], "apm": [ { "ref_id": "apm-ref-id", "elasticsearch_cluster_ref_id": "es-ref-id", "region": "ece-region", "plan": { "cluster_topology": [ { "instance_configuration_id": "apm", "size": { "value": 0, "resource": "memory" }, "zone_count": 1 } ], "apm": {} } } ], "enterprise_search": [ { "ref_id": "enterprise_search-ref-id", "elasticsearch_cluster_ref_id": "es-ref-id", "region": "ece-region", "plan": { "cluster_topology": [ { "node_type": { "appserver": true, "worker": true, "connector": true }, "instance_configuration_id": "enterprise.search", "size": { "value": 0, "resource": "memory" }, "zone_count": 2 } ], "enterprise_search": {} } } ] } }, "system_owned": false }'
After the template is updated, you can start creating new deployments or migrating existing ones to node_roles
.
Although node_roles
and autoscaling are only available in more recent Elastic stack versions, an updated template can still be used with deployments that have versions below 7.10.
In these cases, the data tiers and autoscaling features will only take effect once the deployment is upgraded to versions 7.10 and 7.11, respectively.
Migrating a deployment to node_roles
editOnce a custom template is updated with node_roles
, the existing deployments associated with this template can be migrated to node_roles
.
This migration can be done automatically by performing one of the following actions through the UI:
- Enable a warm, cold, or frozen tier.
- Upgrade the deployment.
- Enable autoscaling (only possible if the custom template has support for autoscaling).
If you do not intend to perform any of these actions, the migration can only be done by manually updating the necessary fields in the deployment plan. This can be performed either through the API or using the deployment Advanced edit page.
Using the API:
- Go to the deployment Edit page.
- Get the deployment update payload by clicking Equivalent API request at the bottom of the page.
-
Update the payload by replacing
node_type
withnode_roles
in each Elasticsearch topology element. To know whichnode_roles
to add to each topology element, refer to the custom template example where support fornode_roles
is added. -
Send a
PUT
request with the updated deployment payload to conclude the migration. Check the Update Deployment API documentation for more details.
Using the Advanced edit:
To follow this approach you need to have administrator privileges.
- Go to the deployment Edit page.
- Click Advanced edit at the bottom of the page.
-
Update the Deployment configuration by replacing
node_type
withnode_roles
in each Elasticsearch topology element. To know whichnode_roles
to add to each topology element, refer to the custom template example where support fornode_roles
is added. - Click Save to conclude the migration.
Once a deployment is migrated to node roles, it is not possible to roll back.
After the migration plan has finished, we recommend following the Migrate index allocation filters to node roles guide.
Step 1 of this guide was already accomplished by adding support for node_roles
.
However, performing steps 2, 3, and 4, which involves updating index settings, index templates, and ILM policies, can prevent shard allocation issues caused by conflicting ILM policies.
Example
editThe following is an example of a deployment plan that does not contain node_roles
:
Example deployment plan with node_type
{ "name": "Example deployment", "prune_orphans": true, "metadata": { "system_owned": false, "hidden": false }, "resources": { "elasticsearch": [ { "ref_id": "es-ref-id", "region": "ece-region", "plan": { "tiebreaker_topology": { "memory_per_node": 1024 }, "cluster_topology": [ { "id": "hot_content", "instance_configuration_id": "custom.data", "zone_count": 2, "node_type": { "master": true, "data": true, "ingest": true }, "elasticsearch": { "node_attributes": { "data": "hot" } }, "size": { "value": 8192, "resource": "memory" } }, { "id": "warm", "instance_configuration_id": "data.highstorage", "zone_count": 1, "node_type": { "data": true, "ingest": false, "master": false }, "elasticsearch": { "node_attributes": { "data": "warm" } }, "size": { "resource": "memory", "value": 0 } }, { "id": "coordinating", "zone_count": 1, "instance_configuration_id": "coordinating", "node_type": { "master": false, "data": false, "ingest": true }, "size": { "value": 0, "resource": "memory" } }, { "id": "master", "zone_count": 1, "instance_configuration_id": "custom.master", "node_type": { "master": true, "data": false, "ingest": false }, "size": { "value": 0, "resource": "memory" } }, { "id": "ml", "zone_count": 1, "instance_configuration_id": "custom.ml", "node_type": { "master": false, "data": false, "ingest": false, "ml": true }, "size": { "value": 0, "resource": "memory" } } ], "elasticsearch": { "version": "7.17.0" }, "deployment_template": { "id": "custom-template" } } } ], "kibana": [ { "region": "ece-region", "ref_id": "kibana-ref-id", "elasticsearch_cluster_ref_id": "es-ref-id", "plan": { "cluster_topology": [ { "instance_configuration_id": "kibana", "size": { "value": 1024, "resource": "memory" }, "zone_count": 1, "kibana": {} } ], "kibana": { "version": "7.17.0" } } } ], "apm": [], "enterprise_search": [] } }
After adding support for node_roles
to the example deployment plan, the resulting plan should look similar to the following:
Example deployment plan with node_roles
{ "name": "Example deployment", "prune_orphans": true, "metadata": { "system_owned": false, "hidden": false }, "resources": { "elasticsearch": [ { "ref_id": "es-ref-id", "region": "ece-region", "plan": { "tiebreaker_topology": { "memory_per_node": 1024 }, "cluster_topology": [ { "id": "hot_content", "instance_configuration_id": "custom.data", "zone_count": 2, "node_roles": [ "master", "ingest", "data_hot", "data_content", "remote_cluster_client", "transform" ], "elasticsearch": { "node_attributes": { "data": "hot" } }, "size": { "value": 8192, "resource": "memory" } }, { "id": "warm", "instance_configuration_id": "data.highstorage", "zone_count": 1, "node_roles": [ "data_warm", "remote_cluster_client" ], "elasticsearch": { "node_attributes": { "data": "warm" } }, "size": { "resource": "memory", "value": 0 } }, { "id": "coordinating", "zone_count": 1, "instance_configuration_id": "coordinating", "node_roles": [ "ingest", "remote_cluster_client" ], "size": { "value": 0, "resource": "memory" } }, { "id": "master", "zone_count": 1, "instance_configuration_id": "custom.master", "node_roles": [ "master", "remote_cluster_client" ], "size": { "value": 0, "resource": "memory" } }, { "id": "ml", "zone_count": 1, "instance_configuration_id": "custom.ml", "node_roles": [ "ml", "remote_cluster_client" ], "size": { "value": 0, "resource": "memory" } } ], "elasticsearch": { "version": "7.17.0" }, "deployment_template": { "id": "custom-template" } } } ], "kibana": [ { "region": "ece-region", "ref_id": "kibana-ref-id", "elasticsearch_cluster_ref_id": "es-ref-id", "plan": { "cluster_topology": [ { "instance_configuration_id": "kibana", "size": { "value": 1024, "resource": "memory" }, "zone_count": 1, "kibana": {} } ], "kibana": { "version": "7.17.0" } } } ], "apm": [], "enterprise_search": [] } }