Use jenkinsci/ec2-fleet-plugin instead of awslabs/ec2-spot-jenkins-plugin
The EC2 Spot Jenkins plugin launches EC2 Spot instances as worker nodes for Jenkins CI server, automatically scaling the capacity with the load.
This plugin uses Spot Fleet to launch instances instead of directly launching them by itself. Amazon EC2 attempts to maintain your Spot fleet's target capacity as Spot prices change to maintain the fleet within the specified price range. For more information, see How Spot Fleet Works.
This plugin is using SemVersion which means that each plugin version looks like
<major>.<minor>.<bugfix>
major = increase only if non back compatible changes
minor = increase when new features
bugfix = increase when bug fixes
As result you safe to update plugin to any version until first number is the same with what you have.
https://github.com/jenkinsci/ec2-fleet-plugin/releases
Specify programmatic access
during creation, and record credentials
which will be used by Jenkins EC2 Fleet Plugin to connect to your Spot Fleet
Add inline policy to the user to allow it use EC2 Spot Fleet AWS documentation about that
{
"Version": "2012-10-17",
"Statement": [
{
"Effect": "Allow",
"Action": [
"ec2:*"
],
"Resource": "*"
},
{
"Effect": "Allow",
"Action": [
"iam:ListRoles",
"iam:PassRole",
"iam:ListInstanceProfiles"
],
"Resource": "*"
}
]
}
https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/spot-fleet-requests.html#create-spot-fleet
Make sure that you:
- Checked
Maintain target capacity
why - specify an SSH key that will be used later by Jenkins.
Once the fleet is launched, you can set it up by adding a new EC2 Fleet cloud in the Jenkins
- Goto
Manage Jenkins > Plugin Manager
- Install
EC2 Fleet Jenkins Plugin
- Goto
Manage Jenkins > Configure System
- Click
Add a new cloud
and selectAmazon SpotFleet
- Configure credentials and specify EC2 Spot Fleet which you want to use
You can specify the scaling limits in your cloud settings. By default, Jenkins will try to scale fleet up if there are enough tasks waiting in the build queue and scale down idle nodes after a specified idleness period.
You can use the History tab in the AWS console to view the scaling history.
Below Groovy script to setup EC2 Spot Fleet Plugin for Jenkins and configure it, you can run it by Jenkins Script Console
import com.amazonaws.services.ec2.model.InstanceType
import com.cloudbees.jenkins.plugins.sshcredentials.impl.BasicSSHUserPrivateKey.DirectEntryPrivateKeySource
import com.cloudbees.jenkins.plugins.sshcredentials.impl.BasicSSHUserPrivateKey
import com.cloudbees.jenkins.plugins.awscredentials.AWSCredentialsImpl
import hudson.plugins.sshslaves.SSHConnector
import hudson.plugins.sshslaves.verifiers.NonVerifyingKeyVerificationStrategy
import com.cloudbees.plugins.credentials.*
import com.cloudbees.plugins.credentials.domains.Domain
import hudson.model.*
import com.amazon.jenkins.ec2fleet.EC2FleetCloud
import jenkins.model.Jenkins
// just modify this config other code just logic
config = [
region: "us-east-1",
fleetId: "...",
idleMinutes: 10,
minSize: 0,
maxSize: 10,
numExecutors: 1,
awsKeyId: "...",
secretKey: "...",
ec2PrivateKey: '''-----BEGIN RSA PRIVATE KEY-----
...
-----END RSA PRIVATE KEY-----'''
]
// https://github.com/jenkinsci/aws-credentials-plugin/blob/aws-credentials-1.23/src/main/java/com/cloudbees/jenkins/plugins/awscredentials/AWSCredentialsImpl.java
AWSCredentialsImpl awsCredentials = new AWSCredentialsImpl(
CredentialsScope.GLOBAL,
"aws-credentials",
config.awsKeyId,
config.secretKey,
"my aws credentials"
)
BasicSSHUserPrivateKey instanceCredentials = new BasicSSHUserPrivateKey(
CredentialsScope.GLOBAL,
"instance-ssh-key",
"ec2-user",
new DirectEntryPrivateKeySource(config.ec2PrivateKey),
"",
"my private key to ssh ec2 for jenkins"
)
// find detailed information about parameters on plugin config page or
// https://github.com/jenkinsci/ec2-fleet-plugin/blob/master/src/main/java/com/amazon/jenkins/ec2fleet/EC2FleetCloud.java
EC2FleetCloud ec2FleetCloud = new EC2FleetCloud(
"", // fleetCloudName
awsCredentials.id,
"",
config.region,
config.fleetId,
"ec2-fleet", // labels
"", // fs root
new SSHConnector(22,
instanceCredentials.id, "", "", "", "", null, 0, 0,
// consult doc for line below, this one say no host verification, but you can use more strict mode
// https://github.com/jenkinsci/ssh-slaves-plugin/blob/master/src/main/java/hudson/plugins/sshslaves/verifiers/NonVerifyingKeyVerificationStrategy.java
new NonVerifyingKeyVerificationStrategy()),
false, // if need to use privateIpUsed
false, // if need alwaysReconnect
config.idleMinutes, // if need to allow downscale set > 0 in min
config.minSize, // minSize
config.maxSize, // maxSize
config.numExecutors, // numExecutors
false, // addNodeOnlyIfRunning
false, // restrictUsage allow execute only jobs with proper label
)
// get Jenkins instance
Jenkins jenkins = Jenkins.getInstance()
// get credentials domain
def domain = Domain.global()
// get credentials store
def store = jenkins.getExtensionList('com.cloudbees.plugins.credentials.SystemCredentialsProvider')[0].getStore()
// add credential to store
store.addCredentials(domain, awsCredentials)
store.addCredentials(domain, instanceCredentials)
// add cloud configuration to Jenkins
jenkins.clouds.add(ec2FleetCloud)
// save current Jenkins state to disk
jenkins.save()
Sometimes you need to prepare slave (which is EC2 instance) before Jenkins could use it. For example install some software which will be required by your builds like Maven etc.
For those cases you have a few options, described below:
Greate for static preconfiguration
AMI allows you to create custom images for your EC2 instances. For example you can create image with Linux plus Java, Maven etc. as result when EC2 fleet will launch new EC2 instance with this AMI it will automatically get all required software. Nice =)
- Create custom AMI as described here
- Create EC2 Spot Fleet with this AMI
EC2 instance allows to specify special script User Data which will be executed when EC2 instance is created. That's allow you to do some customization for particular instance.
However, EC2 instance doesn't provide any information about User Data execution status, as result Jenkins could start task on new instances while User Data still in progress.
To avoid that you can use Jenkins SSH Launcher Prefix Start Agent Command
setting
to specify command which should fail if User Data is not finished, in that way Jenkins will
not be able to connect to instance until User Data is not done more
- Prepare User Data script
- Open Jenkins
- Goto
Manage Jenkins > Configure Jenkins
- Find proper fleet configuration and click
Advance
for SSH Launcher - Add checking command into field
Prefix Start Slave Command
- example
java -version &&
- example
- To apply for existent instances restart Jenkins or Delete Nodes from Jenkins so they will be reconnected
Plugin usage statistic per Jenkins version here
https://jenkins.io/doc/developer/publishing/releasing/
mvn release:prepare release:perform
https://issues.jenkins-ci.org/browse/JENKINS-53954
sudo yum install java-1.8.0
sudo yum remove java-1.7.0-openjdk
java -version