8.4.0-alpha.1

This prerelease introduces a new typed API generated from the elasticsearch-specification. This generation from the common specification allows us to provide a complete API which uses an exhaustive hierarchy of types reflecting the possibilities given by Elasticsearch.

This new API is the next iteration of the Go client for Elasticsearch, it now lives alongside the existing API, it is in alpha state and will gain features over time and releases.

What's new

The TypedClient is built around a fluent builder for easier request creation and a collection of structures and helpers that mimics as closely as possible the Elasticsearch JSON API.

As a first example, here is a search request:

cfg := elasticsearch.Config{
	// Define your configuration
}
es, _ := elasticsearch.NewTypedClient(cfg)
res, err := es.Search().
    Index("index_name").
    Request(&search.Request{
        Query: &types.QueryContainer{
            Match: map[types.Field]types.MatchQuery{
                "name": {Query: "Foo"},
            },
        },
    },
    ).Do(context.Background())

The Request uses the structures found in the typedapi/types package which will lead you along the possibilities. A builder for each structure that allows easier access and declaration is also provided.

More on the specifics and a few examples of standard use-cases can be found in the TypedAPI section of the documentation.

Limitations

While most of the endpoints are covered, a few points are still being worked on and will be part of future releases:

• NDJSON endpoints: bulk, msearch, msearch_template, ML.post_data, find_structure, to name a few.
• Response and Errors structures with deserialization.

Transport & config

While being different, the new API uses all the existing layers that were built so far, elastic-transport-go remains the preferred transport and all your configuration and credentials applies, same as before.

Feedback

Feedback is very welcome, play with it, use it, let us know what you think!

8.3.0

API

• ML.InferTrainedModelDeployment renamed to InferTrainedModel
• ML.PreviewDatafeed has two new parameters, start and end. Documentation
• ML.StartTrainedModelDeployment has three new parameters, number_of_allocations, threads_per_allocation and queue_capacity. Documentation
• Cluster.DeleteVotingConfigExclusions has a new master_timeout parameter.
• Cluster.PostVotingConfigExclusions has a new master_timeout parameter.
• Snapshot.Get has a new index_names parameters (boolean). Whether to include the name of each index in the snapshot. Defaults to true.

New APIs

• Security.HasPrivilegesUserProfile (Experimental API) Documentation

8.2.0

Client

• Fixed a serialisation error for retry_on_conflict in the BulkIndexer. Thanks to @lpflpf for the help!
• Fixed a concurrent map error in the BulkIndexer when custom headers are applied. Thanks to @chzhuo for the contribution!

API

New APIs

• Cat.ComponentTemplates

• ML.GetMemoryStats documentation

• Security.activateUserProfile (Experimental API) Documentation

• Security.disableUserProfile (Experimental API) Documentation

• Security.enableUserProfile (Experimental API) Documentation

• Security.getUserProfile (Experimental API) Documentation

• Security.suggestUserProfiles (Experimental API) Documentation

• Security.updateUserProfileData (Experimental API) Documentation

8.1.0

API

• API is generated from the Elasticsearch 8.1.0 specification.

New parameters

• WithWaitForCompletion for Indices.Forcemerge
• WithFeatures for Indices.Get
• WithForce for ML.DeleteTrainedModel

New APIs

• OidcAuthenticate, OidcLogout and OidcPrepareAuthentication see documentation
• TransformResetTransform

7.17.1

Client

• Fixed serialization of the routing parameter for the BulkIndexerItem routing property.

8.0.0

Client

• The client now uses elastic-transport-go dependency which lives in its own repository.
• With the knewly extracted transport, the retryOnTimeout has been replaced with a retryOnError callback. This allows to select more finely which error should be retried by the client.
• BulkIndexerItem Body field is now an io.ReadSeeker allowing reread without increasing memory consumption.
• BulkIndexerItem know correctly uses the routing property instead of the deprecated _routing.

API

• API is generated from the Elasticsearch 8.0.0 specification.

7.17.0

Client

• Fixed a race condition on metrics in transport #397, thanks @mainliufeng !
• The client will now return an error if a required arguments is passed as a nil value. #201
• API is compatible with Elasticsearch 7.17.0

8.0.0-alpha

Client

Elastic Transport

• This is the first release using the all new elastictransport which now lives in the elastic-transport-go repository. The goal is to allow for reuse between this and future Go clients.

API

• Changed the Body parameter for BulkIndexerItem in favor of io.ReadSeeker to lower memory consumption when using the esutil.BulkIndexer.
• Replaced the Config option RetryOnTimeout in favor of a new RetryOnError function which allows to dynamically chose what error should be retried.

7.16.0

Client

• Adds versioning and routing options to the BulkIndexer. Thanks to @mehran-prs and @munkyboy !

• Adds CA fingerprinting. You can configure the client to only trust certificates that are signed by a specific CA certificate (CA certificate pinning) by providing a ca_fingerprint option. This will verify that the fingerprint of the CA certificate that has signed the certificate of the server matches the supplied value:

    elasticsearch.NewClient(Config{
        Addresses:              []string{"https://elastic:changeme@localhost:9200"},
        CertificateFingerprint: "A6FB224A4386..."
    })

API

• New APIs:

  • ModifyDataStream, documentation
  • TransformUpgradeTransforms, documentation
  • Migration.GetFeatureUpgradeStatus, documentation
  • Migration.PostFeatureUpgrade, documentation

• New Experimental API:

  • FleetSearch, documentation
  • FleetMsearch, documentation

• Changes:

  • DeleteByQuery
    • Removed _source, _source_excludes, _source_includes parameters.
  • UpdateByQuery
    • Removed _source, _source_excludes, _source_includes parameters.
  • OpenPointInTime
    • The parameter keep_alive is now required.
  • SearchMvt
    • Added TrackTotalHits, Indicate if the number of documents that match the query should be tracked. A number can also be specified, to accurately track the total hit count up to the number.
  • IngestPutPipeline
    • Added WithIfVersion, required version for optimistic concurrency control for pipeline updates.
  • IndicesGetIndexTemplate
    • WithName, a pattern that returned template names must match.
  • NodesHotThreads
    • WithSort, the sort order for 'cpu' type (default: total).
  • MLPutTrainedModel
    • WithDeferDefinitionDecompression, if set to true and a compressed_definition is provided, the request defers definition decompression and skips relevant validations.
  • TransformDeleteTransform
    • WithTimeout, controls the time to wait for the transform deletion.
  • TransformPutTransform
    • WithTimeout, controls the time to wait for the transform to start.
  • TransformUpdateTransform
    • WithTimeout, controls the time to wait for the update.

• Promoted to stable:

  • FleetGlobalCheckpoints
  • GetScriptContext
  • GetScriptLanguages
  • IndicesResolveIndex
  • MonitoringBulk
  • RankEval
  • SearchableSnapshotsMount
  • SearchableSnapshotsStats
  • SecurityClearCachedServiceTokens
  • SecurityCreateServiceToken
  • SecurityDeleteServiceToken
  • SecurityGetServiceAccounts
  • SecurityGetServiceCredentials
  • ShutdownDeleteNode
  • ShutdownGetNode
  • ShutdownPutNode
  • TermsEnum

7.15.1

Client

• Allow User-Agent override via the userAgentHeader header. Credit goes to @aleksmaus!