Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[Backport 2.x] [DOC] Configure the Spark metrics properties while creating a s3 Glue Connector #2508

Merged
merged 1 commit into from
Feb 6, 2024
Merged

[Backport 2.x] [DOC] Configure the Spark metrics properties while creating a s3 Glue Connector #2508

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
56 changes: 42 additions & 14 deletions docs/user/interfaces/asyncqueryinterface.rst
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -20,21 +20,49 @@ All the queries to be executed on spark execution engine can only be submitted v

Required Spark Execution Engine Config for Async Query APIs
===========================================================
Currently, we only support AWS EMRServerless as SPARK execution engine. The details of execution engine should be configured under
``plugins.query.executionengine.spark.config`` in cluster settings. The value should be a stringified json comprising of ``applicationId``, ``executionRoleARN``,``region``, ``sparkSubmitParameter``.
Sample Setting Value ::

plugins.query.executionengine.spark.config:
'{ "applicationId":"xxxxx",
"executionRoleARN":"arn:aws:iam::***********:role/emr-job-execution-role",
"region":"eu-west-1",
"sparkSubmitParameter": "--conf spark.dynamicAllocation.enabled=false"
}'
The user must be careful before transitioning to a new application or region, as changing these parameters might lead to failures in the retrieval of results from previous async query jobs.
The system relies on the default AWS credentials chain for making calls to the EMR serverless application. It is essential to confirm that the default credentials possess the necessary permissions to pass the role required for EMR job execution, as specified in the engine configuration.
* ``applicationId``, ``executionRoleARN`` and ``region`` are required parameters.
* ``sparkSubmitParameter`` is an optional parameter. It can take the form ``--conf A=1 --conf B=2 ...``.
Currently, the system supports only AWS EMRServerless as the SPARK execution engine. Configuration details for the execution engine should be specified under ``plugins.query.executionengine.spark.config`` in the opensearch.yml or cluster settings. The configuration value is expected to be a JSON string that includes ``applicationId``, ``executionRoleARN``, ``region``, and ``sparkSubmitParameter``.

Sample Setting Value in opensearch.yml
--------------------

.. code-block:: yaml

"plugins.query.executionengine.spark.config: '{\"applicationId\":\"xxxxx\",\"executionRoleARN\":\"arn:aws:iam::xxxxx:role/emr-job-execution-role\",\"region\":\"us-west-2\", \"sparkSubmitParameters\": \"--conf spark.dynamicAllocation.enabled=false\"}'"

Caution
-------

Users must exercise caution when transitioning to a new application or region, as changes to these parameters may lead to failures in retrieving results from previous asynchronous query jobs.

The system utilizes the default AWS credentials chain for calls to the EMR serverless application. It is critical to ensure that the default credentials have the necessary permissions to assume the role required for EMR job execution, as delineated in the engine configuration.

Requirements
-------------

- **Required Parameters**: ``applicationId``, ``executionRoleARN``, and ``region`` must be provided.
- **Optional Parameter**: ``sparkSubmitParameter`` is optional and can be formatted as ``--conf A=1 --conf B=2 ...``.

AWS CloudWatch metrics configuration
-------------

Starting with Flint 0.1.1, users can utilize AWS CloudWatch as an external metrics sink while configuring their own metric sources. Below is an example of a console request for setting this up:

.. code-block:: json

PUT _cluster/settings
{
"persistent": {
"plugins.query.executionengine.spark.config": "{\"applicationId\":\"xxxxx\",\"executionRoleARN\":\"arn:aws:iam::xxxxx:role/emr-job-execution-role\",\"region\":\"us-east-1\",\"sparkSubmitParameters\":\"--conf spark.dynamicAllocation.enabled=false --conf spark.metrics.conf.*.sink.cloudwatch.class=org.apache.spark.metrics.sink.CloudWatchSink --conf spark.metrics.conf.*.sink.cloudwatch.namespace=OpenSearchSQLSpark --conf spark.metrics.conf.*.sink.cloudwatch.regex=(opensearch|numberAllExecutors).* --conf spark.metrics.conf.*.source.cloudwatch.class=org.apache.spark.metrics.source.FlintMetricSource \"}"
}
}

For a comprehensive list of Spark configuration options related to metrics, please refer to the Spark documentation on monitoring:

- Spark Monitoring Documentation: https://spark.apache.org/docs/latest/monitoring.html#metrics

Additionally, for details on setting up CloudWatch metric sink and Flint metric source, consult the OpenSearch Spark project:

- OpenSearch Spark GitHub Repository: https://github.com/opensearch-project/opensearch-spark

Async Query Creation API
======================================
Expand Down
Loading