Azure repository

edit

You can use Azure Blob storage as a repository for Snapshot and restore.

Setup

edit

To enable Azure repositories, first configure an Azure repository client by specifying one or more settings of the form azure.client.CLIENT_NAME.SETTING_NAME. By default, azure repositories use a client named default, but you may specify a different client name when registering each repository.

The only mandatory Azure repository client setting is account, which is a secure setting defined in the Elasticsearch keystore. To provide this setting, use the elasticsearch-keystore tool on each node:

bin/elasticsearch-keystore add azure.client.default.account

If you adjust this setting after a node has started, call the Nodes reload secure settings API to reload the new value.

You may define more than one client by setting their account values. For instance, to set the default client and another client called secondary, run the following commands on each node:

bin/elasticsearch-keystore add azure.client.default.account
bin/elasticsearch-keystore add azure.client.secondary.account

The key and sas_token settings are also secure settings and can be set using commands like the following:

bin/elasticsearch-keystore add azure.client.default.key
bin/elasticsearch-keystore add azure.client.secondary.sas_token

Other Azure repository client settings must be set in elasticsearch.yml before the node starts. For example:

azure.client.default.timeout: 10s
azure.client.default.max_retries: 7
azure.client.default.endpoint_suffix: core.chinacloudapi.cn
azure.client.secondary.timeout: 30s

In this example, the client side timeout is 10s per try for repositories which use the default client, with 7 retries before failing and an endpoint suffix of core.chinacloudapi.cn. Repositories which use the secondary client will have a timeout of 30s per try, but will use the default endpoint and will fail after the default number of retries.

Once an Azure repository client is configured correctly, register an Azure repository as follows, providing the client name using the client repository setting:

resp = client.snapshot.create_repository(
    name="my_backup",
    repository={
        "type": "azure",
        "settings": {
            "client": "secondary"
        }
    },
)
print(resp)
const response = await client.snapshot.createRepository({
  name: "my_backup",
  repository: {
    type: "azure",
    settings: {
      client: "secondary",
    },
  },
});
console.log(response);
PUT _snapshot/my_backup
{
  "type": "azure",
  "settings": {
    "client": "secondary"
  }
}

If you are using the default client, you may omit the client repository setting:

resp = client.snapshot.create_repository(
    name="my_backup",
    repository={
        "type": "azure"
    },
)
print(resp)
const response = await client.snapshot.createRepository({
  name: "my_backup",
  repository: {
    type: "azure",
  },
});
console.log(response);
PUT _snapshot/my_backup
{
  "type": "azure"
}

In progress snapshot or restore jobs will not be preempted by a reload of the storage secure settings. They will complete using the client as it was built when the operation started.

Client settings

edit

The following list describes the available client settings. Those that must be stored in the keystore are marked as (Secure, reloadable); the other settings must be stored in the elasticsearch.yml file. The default CLIENT_NAME is default but you may configure a client with a different name and specify that client by name when registering a repository.

azure.client.CLIENT_NAME.account (Secure, reloadable)
The Azure account name, which is used by the repository’s internal Azure client. This setting is required for all clients.
azure.client.CLIENT_NAME.endpoint_suffix
The Azure endpoint suffix to connect to. The default value is core.windows.net.
azure.client.CLIENT_NAME.key (Secure, reloadable)
The Azure secret key, which is used by the repository’s internal Azure client. Alternatively, use sas_token.
azure.client.CLIENT_NAME.max_retries
The number of retries to use when an Azure request fails. This setting helps control the exponential backoff policy. It specifies the number of retries that must occur before the snapshot fails. The default value is 3. The initial backoff period is defined by Azure SDK as 30s. Thus there is 30s of wait time before retrying after a first timeout or failure. The maximum backoff period is defined by Azure SDK as 90s.
azure.client.CLIENT_NAME.proxy.host
The host name of a proxy to connect to Azure through. By default, no proxy is used.
azure.client.CLIENT_NAME.proxy.port
The port of a proxy to connect to Azure through. By default, no proxy is used.
azure.client.CLIENT_NAME.proxy.type
Register a proxy type for the client. Supported values are direct, http, and socks. For example: azure.client.default.proxy.type: http. When proxy.type is set to http or socks, proxy.host and proxy.port must also be provided. The default value is direct.
azure.client.CLIENT_NAME.sas_token (Secure, reloadable)
A shared access signatures (SAS) token, which the repository’s internal Azure client uses for authentication. The SAS token must have read (r), write (w), list (l), and delete (d) permissions for the repository base path and all its contents. These permissions must be granted for the blob service (b) and apply to resource types service (s), container (c), and object (o). Alternatively, use key.
azure.client.CLIENT_NAME.timeout
The client side timeout for any single request to Azure, as a time unit. For example, a value of 5s specifies a 5 second timeout. There is no default value, which means that Elasticsearch uses the default value set by the Azure client.
azure.client.CLIENT_NAME.endpoint
The Azure endpoint to connect to. It must include the protocol used to connect to Azure.
azure.client.CLIENT_NAME.secondary_endpoint
The Azure secondary endpoint to connect to. It must include the protocol used to connect to Azure.

Obtaining credentials from the environment

If you specify neither the key nor the sas_token settings for a client then Elasticsearch will attempt to automatically obtain credentials from the environment in which it is running using mechanisms built into the Azure SDK. This is ideal for when running Elasticsearch on the Azure platform.

When running Elasticsearch on an Azure Virtual Machine, you should use Azure Managed Identity to provide credentials to Elasticsearch. To use Azure Managed Identity, assign a suitably authorized identity to the Azure Virtual Machine on which Elasticsearch is running.

When running Elasticsearch in Azure Kubernetes Service, for instance using Elastic Cloud on Kubernetes, you should use Azure Workload Identity to provide credentials to Elasticsearch. To use Azure Workload Identity, mount the azure-identity-token volume as a subdirectory of the Elasticsearch config directory and set the AZURE_FEDERATED_TOKEN_FILE environment variable to point to a file called azure-identity-token within the mounted volume.

The Azure SDK has several other mechanisms to automatically obtain credentials from its environment, but the two methods described above are the only ones that are tested and supported for use in Elasticsearch.

Repository settings

edit

The Azure repository supports the following settings, which may be specified when registering an Azure repository as follows:

resp = client.snapshot.create_repository(
    name="my_backup",
    repository={
        "type": "azure",
        "settings": {
            "client": "secondary",
            "container": "my_container",
            "base_path": "snapshots_prefix"
        }
    },
)
print(resp)
const response = await client.snapshot.createRepository({
  name: "my_backup",
  repository: {
    type: "azure",
    settings: {
      client: "secondary",
      container: "my_container",
      base_path: "snapshots_prefix",
    },
  },
});
console.log(response);
PUT _snapshot/my_backup
{
  "type": "azure",
  "settings": {
    "client": "secondary",
    "container": "my_container",
    "base_path": "snapshots_prefix"
  }
}
client
The name of the Azure repository client to use. Defaults to default.
container
Container name. You must create the azure container before creating the repository. Defaults to elasticsearch-snapshots.
base_path

Specifies the path within container to repository data. Defaults to empty (root directory).

Don’t set base_path when configuring a snapshot repository for Elastic Cloud Enterprise. Elastic Cloud Enterprise automatically generates the base_path for each deployment so that multiple deployments may share the same bucket.

chunk_size
Big files can be broken down into multiple smaller blobs in the blob store during snapshotting. It is not recommended to change this value from its default unless there is an explicit reason for limiting the size of blobs in the repository. Setting a value lower than the default can result in an increased number of API calls to the Azure blob store during snapshot create as well as restore operations compared to using the default value and thus make both operations slower as well as more costly. Specify the chunk size as a byte unit, for example: 10MB, 5KB, 500B. Defaults to the maximum size of a blob in the Azure blob store which is 5TB.
compress
When set to true metadata files are stored in compressed format. This setting doesn’t affect index files that are already compressed by default. Defaults to true.
max_restore_bytes_per_sec
(Optional, byte value) Maximum snapshot restore rate per node. Defaults to unlimited. Note that restores are also throttled through recovery settings.
max_snapshot_bytes_per_sec
(Optional, byte value) Maximum snapshot creation rate per node. Defaults to 40mb per second. Note that if the recovery settings for managed services are set, then it defaults to unlimited, and the rate is additionally throttled through recovery settings.
readonly

(Optional, Boolean) If true, the repository is read-only. The cluster can retrieve and restore snapshots from the repository but not write to the repository or create snapshots in it.

Only a cluster with write access can create snapshots in the repository. All other clusters connected to the repository should have the readonly parameter set to true.

If false, the cluster can write to the repository and create snapshots in it. Defaults to false.

If you register the same snapshot repository with multiple clusters, only one cluster should have write access to the repository. Having multiple clusters write to the repository at the same time risks corrupting the contents of the repository.

location_mode
primary_only or secondary_only. Defaults to primary_only. Note that if you set it to secondary_only, it will force readonly to true.

Repository validation rules

edit

According to the containers naming guide, a container name must be a valid DNS name, conforming to the following naming rules:

  • Container names must start with a letter or number, and can contain only letters, numbers, and the dash (-) character.
  • Every dash (-) character must be immediately preceded and followed by a letter or number; consecutive dashes are not permitted in container names.
  • All letters in a container name must be lowercase.
  • Container names must be from 3 through 63 characters long.

Supported Azure Storage Account types

The Azure repository type works with all Standard storage accounts

  • Standard Locally Redundant Storage - Standard_LRS
  • Standard Zone-Redundant Storage - Standard_ZRS
  • Standard Geo-Redundant Storage - Standard_GRS
  • Standard Read Access Geo-Redundant Storage - Standard_RAGRS

Premium Locally Redundant Storage (Premium_LRS) is not supported as it is only usable as VM disk storage, not as general storage.

Linearizable register implementation

edit

The linearizable register implementation for Azure repositories is based on Azure’s support for strongly consistent leases. Each lease may only be held by a single node at any time. The node presents its lease when performing a read or write operation on a protected blob. Lease-protected operations fail if the lease is invalid or expired. To perform a compare-and-exchange operation on a register, Elasticsearch first obtains a lease on the blob, then reads the blob contents under the lease, and finally uploads the updated blob under the same lease. This process ensures that the read and write operations happen atomically.