v1.3.0

feat: renamed apiserver to httpserver; (#48)

v1.2.6

What's Changed

• fix: revert cpu and memory average empty space by @mfroembgen in #44

Full Changelog: v1.2.5...v1.2.6

v1.2.5

What's Changed

• fix: cpu and memory average empty space by @mfroembgen in #42

New Contributors

• @mfroembgen made their first contribution in #42

Full Changelog: v1.2.4...v1.2.5

v1.2.4

What's Changed

• fix: remove double quotes from label formatting by @ztzxt in #41

New Contributors

• @ztzxt made their first contribution in #41

Full Changelog: v1.2.3...v1.2.4

v1.2.3

What's Changed

• fix: traefik related request count per target query by @applike-ss in #39

Full Changelog: v1.2.2...v1.2.3

v1.2.2

What's Changed

• Fix traefik related metrics by @applike-ss in #38

Full Changelog: v1.2.1...v1.2.2

v1.2.1

What's Changed

• fix: add traefik panel only when api routes exist by @applike-ss in #34

Full Changelog: v1.2.0...v1.2.1

v1.2.0

This release just updates github workflows.

For changes since v1.0.0, please check out Release v1.0.0

v1.1.0

In this release i have added a few new parameters required for provider setup.

BREAKING CHANGES

New provider attribute orchestrator

To make this provider kubernetes compatible with our ingress router of choice (traefik), i added the new orchestrator attribute to the provider which has these allowed values: ecs, kubernetes.

Setting this to ecs will keep the original behavior by querying CloudWatch for EC2 Load Balancer metrics, while setting this to kubernetes will make it use different queries for both the application as well as the traefik load balancer.

New sub-attributes for the name_patterns attribute

The provider now needs to be provided with additional attributes in the name_patterns attribute.
These are the following:

• kubernetes_namespace for a pattern to generate a namespace for prometheus metric queries.
• kubernetes_pod for a pattern to generate a pod name for prometheus metric queries.
  • It is highly suggested not to use wildcards here.
  • The provider will attach wildcards as needed for metrics labels. Say you have a pod foo-75f5f74749-62p6c (deployment) or foo-0 (statefulset), then adjust the pattern to foo.
• traefik_service_name for a pattern to generate the traefik service name as in the traefik metric labels. This usually goes like namespace-serviceName-portNumber@kubernetes.

v1.0.0

In this release i have added the handling of the new gosoline metadata server structure changes.

BREAKING CHANGES

This release requires the use of gosoline v0.7.0 or newer due to changes in the metadata server output structure

Adjustments on existing implementations

SQS queues are not a simple list of strings anymore, but are objects with these properties:

• aws_client_name
• queue_arn
• queue_name
• queue_name_full
• queue_url

DynamoDB tables are not a simple list of strings anymore, but are objects with these properties:

• aws_client_name
• table_name

The provider now needs to be provided with an additional property in the name_patterns input, called grafana_cloudwatch_datasource (defaults to cloudwatch-{family} otherwise).
For CloudWatch related datasources we do not rely on the Grafana default CloudWatch datasource called CloudWatch anymore.
This is to allow multi-account usage inside a single Grafana instance.
As a side-effect we are now using the aws_client_name supplied by the given resources in the gosoline_application_dashboard_definition datasource to determine the Grafana datasource.
Whenever the default client is used (called default), we instead use the provider configured CloudWatch datasource.

This is also true for the following panels:

• API Server
• ELB
• Error/Warn
• Stream consumers
• Stream producers

NON-BREAKING CHANGES

Additional properties on existing implementations

kinsumers now have these additional properties:

• aws_client_name
• open_shard_count
• stream_arn

kinesis record writers now have these additional properties:

• aws_client_name
• open_shard_count
• stream_arn