HOW TO: Check for Elasticsearch / Kinesis Stream Metrics


Overview / Description

Refer to this article for steps in checking Elasticsearch Health in the AWS console.


AWS Elasticsearch Domain Metrics for eMite

1. Log in to the AWS Console. You’ll need an account with access permissions to AWS Elasticsearch.

2. Change the region to the region where the AWS Elasticsearch domain is deployed. This would be the same region where your Amazon Connect instance is deployed.

3. Select the Elasticsearch domain used by eMite.

4. Select Cluster Health tab.

Recommended metrics for monitoring:

METRIC

DESCRIPTION

METRIC

DESCRIPTION

Cluster Status

General indicative health of the cluster:

  • Green - All primary and secondary shards are allocated and accessible

  • Yellow - All primary shards are allocated

  • Red - Some primary shards are not accessible/corrupted

    • When cluster status is Red, eMite application will not be able to read/write from Elasticsearch

    • Red status generally requires the affected indices to be restored from snapshot

Cluster Write Status

  • Red - Elasticsearch is not accepting write requests

    • Usually caused by elasticsearch running out of free storage space, or a Red cluster status

Snapshot Failure Status

  • Red - Hourly snapshot is failing

    • Usually caused by a Red Cluster status

Total Free Storage Space

  • Indicates amount of free storage space left

  • If elasticsearch runs out of storage space, write requests will be rejected – no new data will be written to elasticsearch

  • Can be fixed by increasing the allocated storage space to the domain

Maximum CPU Utilization

  • High CPU utilization may be caused by more agent activity (resulting in more data being written to elasticsearch) or more concurrent active eMite dashboards

  • High CPU utilization can be resolved by increasing the size of the elasticsearch domain data node (e.g. increase from m5.large to m5.xlarge)

Maximum JVM Memory Pressure

  • Normal utilization pattern is sawtooth pattern going up to around 80% and then going down

  • Utilization > 80% may be caused by more data stored on elasticsearch, or more agent activity

  • High Memory utilization can be resolved by increasing the size of the elasticsearch domain data node (e.g. increase from m5.large to r5.large)


AWS Kinesis Stream Metrics for eMite

1. Log in to the AWS Console. You’ll need an account with access permissions to AWS Kinesis.

2. Change the region to the region where the Kinesis Streams are deployed. This would be the same region where your Amazon Connect instance is deployed.

3. Select the Agent Events or CTR Kinesis Stream.

4. Select Monitoring tab.

Recommended metrics for monitoring:

METRIC

DESCRIPTION

METRIC

DESCRIPTION

Write throughput exceeded - average (Percent)

  • Value greater than 0 - PUT request to the stream was throttled, which could indicate the data input to the stream exceeds the current capacity

  • Can be fixed by increasing number of shards for the stream

Read throughput exceeded - average (Percent)

  • Value greater than 0 - GET request to the stream was throttled, which could indicate the data output of the stream exceeds the current capacity

  • Can be fixed by increasing number of shards for the stream