Releases: cloudability/metrics-agent
metrics-agent-2.11.1
Updates agent startup to log connection status of all ready nodes. Determines connection status as a whole rather than simply checking the first node.
metrics-agent-2.11.0
Removes prometheus data collection from the agent.
Removes cadvisor data collection from the agent.
Removes all flags related to these parts of the agent.
AKS K8s 1.23 Support and gopkg.in/yaml.v3 Upgrade
Container Cost Allocation is now officially supported on Kubernetes 1.23 version for Azure Kubernetes Service (AKS). This capability will enable customers to gain visibility into their container resource usages, and the cost for their clusters running on Kubernetes 1.23 version.
Customers should now be able to download and deploy our metrics agent using the regular provisioning workflow.
Note: Our metrics agent has not been tested with Amazon Elastic Kubernetes Service (EKS) 1.23 version or Google Kubernetes Engine (GKE). After AWS & GKE announces support for 1.23 version, we will validate our metrics agent and update our release notes.
This release also includes a Snyk vulnerability fix which updates the package gopkg.in/yaml.v3 from 3.0.0-beta to 3.0.1
metrics-agent-2.10.1
A Helm chart for Kubernetes
Parallel Collection and Sanitization
This release includes 2 new features.
Parallel node collection (enabled by default)
Includes a new feature that allows for node polling parallelism. The metrics-agent polls metrics from each node every polling interval and now this polling is done concurrently. There is a new env variable that can be set called CLOUDABILITY_NUMBER_OF_CONCURRENT_NODE_POLLERS which is the max number of concurrent pollers at a time. The current default is set to 100.
Parsing and sanitization of data from core files (disabled by default)
Adds a flag CLOUDABILITY_PARSE_METRIC_DATA that will parse and remove much non-essential data from the metrics agent. This decreases core file sizes by approximately 15%, but may differ depending on customer cluster configurations.
The performance impact of this on large clusters is currently undetermined, and enabling this flag carries risk of metrics agent slowdowns/crashes. If enabled extra headroom should be given to the agent to determine an appropriate cpu/memory footprint. This is currently flagged off by default
metrics-agent-2.10.0 Helm
This release includes 2 new features.
Parallel node collection (enabled by default)
Includes a new feature that allows for node polling parallelism. The metrics-agent polls metrics from each node every polling interval and now this polling is done concurrently. There is a new env variable that can be set called CLOUDABILITY_NUMBER_OF_CONCURRENT_NODE_POLLERS which is the max number of concurrent pollers at a time. The current default is set to 100.
Parsing and sanitization of data from core files (disabled by default)
Adds a flag CLOUDABILITY_PARSE_METRIC_DATA that will parse and remove much non-essential data from the metrics agent. This decreases core file sizes by approximately 15%, but may differ depending on customer cluster configurations.
The performance impact of this on large clusters is currently undetermined, and enabling this flag carries risk of metrics agent slowdowns/crashes. If enabled extra headroom should be given to the agent to determine an appropriate cpu/memory footprint. This is currently flagged off by default
Service Account Token Refresh Support
The Kubernetes feature "BoundServiceAccountTokenVolume" graduated to beta and is now enabled on 1.21+ clusters which improves security of service account tokens. This release refreshes the service account token that the metrics-agent's clients use every time node metrics are collected. This prevents the metrics-agent from using stale tokens. For more information see the Kubernetes official release statement and EKS documentation.
metrics-agent-2.9 Helm
The Kubernetes feature "BoundServiceAccountTokenVolume" graduated to beta and is now enabled on 1.21+ clusters which improves security of service account tokens. This release refreshes the service account token that the metrics-agent's clients use every time node metrics are collected. This prevents the metrics-agent from using stale tokens. For more information see the Kubernetes official release statement and EKS documentation.
Helm Support for Metrics-agent
Adds supporting for deploying the metrics-agent using Helm
See README for deployment instructions
Helm Support for Metrics-agent
Adds supporting for deploying the metrics-agent using Helm
See README for deployment instructions