This tool provides the metric collection agents for the following New Relic plugins:
- EC2
- EBS
- ELB
- RDS
- SQS
- SNS
- ElastiCache
- Memcached
- Redis
- ASG
- A single t1.micro EC2 instance (in any region)
- Ruby (>= 1.9.2)
- Rubygems (>= 1.3.7)
- Bundler
gem install bundler - Git
- Download the latest tagged version from https://github.com/newrelic-platform/newrelic_aws_cloudwatch_extension/tags
- Extract to the location you want to run the plugin from
- Rename
config/template_newrelic_plugin.ymltoconfig/newrelic_plugin.yml - Edit
config/newrelic_plugin.yml - Run
bundle install - Run
bundle exec ./bin/newrelic_aws(or on Windowsbundle exec ruby bin\newrelic_aws)
Chef and Puppet are tools that automate software installation. The Amazon CloudWatch plugin has installation support for both:
Note: For more information on using Chef and Puppet with New Relic, see the New Relic docs.
This plugin is configured through the config/newrelic_plugin.yml file. It requires:
- a New Relic license key that can be found at https://rpm.newrelic.com/extensions/com.newrelic.aws.ec2
- an AWS Access Key
- an AWS Secret Key
The plugin can also be configured to query specific CloudWatch regions, e.g. us-east-1 or us-west-1. By default the plugin will query all available regions.
Amazon ElastiCache supports both Memcached and Redis caching technologies. The Memcached agent is configured under the ec section, while the Redis agent is configured under the ecr section.
Amazon AutoScaling is a sub-service of EC2, and in CloudWatch is reported in the EC2 namespace. However, the New Relic plugin has been implemented as a seperate plugin, as this allows you to monitoring Auto Scaling groups independently of individual instances.
Filtering instances by tags is supported for both EC2 and EBS. Details on adding tags to instances is available in the AWS documentation.
A list of case-sensitive tags can be added to the yml configuration for EC2 and EBS and only instances containing one or more of those tags will be monitored. This can reduce CloudWatch requests and costs.
Tagged instances will be monitored if they:
- have a tag with a matching key in the tag list (e.g. tagged with key
newrelic_monitoredand any value:true,yes, etc.) - have a
Nameornametag with a matching value in the tag list (e.g. tagged with keyNameand valueprod_1_db)
If there are no configured tags, all available instances will be monitored. This is the default behavior.
...
agents:
ec2:
enabled: true
tags:
- newrelic_monitored
- prod_1_db
ebs:
enabled: true
tags:
- newrelic_monitored
As noted below, there is a default 60 second delay in reporting metrics from CloudWatch which adjusts the time window for queried data. This is due to CloudWatch metrics not being immediately available for querying as they may take some time to process. Unfortunately there is little that can be done from the plugin to address this, besides adjusting the time window for metric querying from CloudWatch. The configuration option cloudwatch_delay can be specified for each AWS agent to override the default 60 second delay.
...
agents:
ec2:
enabled: true
cloudwatch_delay: 120
ebs:
enabled: true
...
####Affected Metrics
RDS/CPU Utilization- If you see 0% CPU utilization, try increasing thecloudwatch_delayoption for the AWS RDS agent.
This plugin is also available as an Amazon Machine Image (AMI) via the AWS Marketplace. Learn more, then quickly install and configure the AMI here: https://aws.amazon.com/marketplace/pp/B00DMMUO0O/
The AMI takes the contents of config/newrelic_plugin.yml as user-data, which is configured when creating the EC2 instance. Once the instance is running with valid user-data, no further action is required. To change the configuration, terminate the current instance and create another.
If you like the AMI, please leave a 5-star review in the AWS Marketplace.
If you don't like the AMI, New Relic would appreciate that you do not leave a bad review on the AWS Marketplace. Instead, open a ticket with New Relic Support and let us know what we could do better. We take your feedback very seriously - and by opening a ticket, we can notify you when we've addressed your feedback.
Note: There is a fantastic blog post on this topic here (screenshots).
This plugin requires AWS API credentials, using IAM is highly recommended, giving it read-only access to select services.
You will need to create a new IAM group, NewRelicCloudWatch, where the permissions will be defined. You will want to use a custom policy for the group, NewRelicCloudWatch, using the following JSON for the policy document.
{
"Version": "2012-10-17",
"Statement": [
{
"Action": [
"autoscaling:Describe*",
"cloudwatch:Describe*",
"cloudwatch:List*",
"cloudwatch:Get*",
"ec2:Describe*",
"ec2:Get*",
"ec2:ReportInstanceStatus",
"elasticache:DescribeCacheClusters",
"elasticloadbalancing:Describe*",
"sqs:GetQueueAttributes",
"sqs:ListQueues",
"rds:DescribeDBInstances",
"SNS:ListTopics"
],
"Effect": "Allow",
"Resource": "*"
}
]
}
To get API credentials, a IAM user must be created, NewRelicCloudWatch. Be sure to save the user access key id and secret access key on creation. Add the user to the NewRelicCloudWatch IAM group. Use the IAM user API credentials in the plugin configuration file.
- CloudWatch detailed monitoring is recommended, please enable it when available. (see Using Amazon CloudWatch section on http://aws.amazon.com/cloudwatch/)
- Chart x-axis (time) is off by 60 seconds, this is due to CloudWatch's lag in reporting metrics.
- Latest data point is used to fill gaps in low resolution metrics.
You can use services like these to manage this process.
The provided daemon.rb file allows newrelic_aws to run as a daemon.
chmod +x bin/daemon
bundle exec bin/daemon start
bundle exec bin/daemon stop
Plugin support and troubleshooting assistance can be obtained by visiting support.newrelic.com
The New Relic AWS plugin was originally authored by Sean Porter and the team at Heavy Water Operations. Subsequent updates and support are provided by New Relic.