Manage AWS Instances

31 Aug, 2023

Recently I have moved into a new position working on the cloud environment in my current company as an AWS Engineer to help build, support, maintain the infrastructure of our systems. With this change has also allowed me to learn more about AWS and use the CLI both locally and on the hosted EC2 instances.

The issue and resolution

There was a time I needed to start a server, but logging into the AWS Console, navigating to the right page, searching, performing the action took a bit of time. Not much, but also was thinking there should be a quicker way to do this from the command line.

Quickly I found the aws ec2 start-instances was the way to go. The problem was I wanted to specify a server name for it to start, and needed the instance id. I found a way to get the instance id from the tag Name and the Value of a server.

(aws ec2 describe-instances --filters "Name=tag:Name,Values=ServerHostName)" --query "Reservations[].Instances[].InstanceId" --profile default --output text --region us-east-1)

This query would describe the instance given the server name and query the JSON output for the InstanceID and output it as text. From here I could then start an EC2 instance

aws ec2 start-instances --instance-ids i-XXXXXXXXXXXX --profile default --output text --region us-east-1

I found a solution, but there is no way I was able to remember the long list of command line switches and JSON filters. So I started writing a function around this idea.

Making the wrapper functions

First I needed to gather the available profiles from the ~/.aws/config file that contained the SSO login and profile information. Next I looped through each line replacing items to get just the profile name. Once that was done, I made a numerical menu listing of the profiles so you can select the profile you'd like to use. If the profile is not listed, you can make a selection then configure the SSO parameters and continue.

While using the CLI I noticed the session token would timeout just like the web console does. For this I also made a wrapper function around the aws call to query the Account information. If your token is still active the length will be 14 and if not, I had it pass along the aws sso login --profile command to refresh our token automatically.

# the call to check if your session is still active or not
(aws sts get-caller-identity --query "Account" --profile $chosenProfile | Select-Object Length).Length

To begin, I wanted to test if my session is active for my profile and if so, proceed. If not, my Select-AWSProfile function will proceed with the aws sso login --profile.

# Read the profiles and send user to select the one they want
if ($null -eq $chosenProfile) {
    $chosenProfile = Select-AWSProfile
}

if (Test-SessionActive -chosenProfile $chosenProfile){
    #Session still active
    if($null -eq $ServerName){
        $ServerName = Read-Host -Prompt "What is the servename"
    }
    Write-Host "Attempting to start $ServerName in $region"
    $instanceID = (aws ec2 describe-instances --filters "Name=tag:Name,Values=$($ServerName)" --query "Reservations[].Instances[].InstanceId" --profile $chosenProfile --output text --region $region)
    aws ec2 start-instances --instance-ids $instanceID --profile $chosenProfile --region $region
}else{
    # Error in the process
    Write-Host "There was an issue running the process: $($MyInvocation.MyCommand)"
}

Part of this process is in the function to pass along a few command line switches: -ServerName, -chosenProfile, -region. All of the wrapper functions have these to pass along or override the defaults for the profiles.

The EC2 management functions are:

Helper functions

After successfully creating the helper functions I found a way to get the instance id of an instance by providing the name of the server. This could be a little off from time to time, but in my case the tag:Name and its value is also the hostname of the machine. This might not always be the case in other situations.

I used the same logic as before but was able to query the JSON for the right function to get the Instance ID from a server name or the server name from the instance id. Both of these functions have an additional switch for -copy to add the output to your clip board if you'd like.

Underlying AWS CLI in the wrapper functions:

# Getting the Server Name from the Instance ID
aws ec2 describe-instances --instance-id $instanceID --profile $chosenProfile --query "Reservations[].Instances[].Tags[?Key=='Name'].Value" --output text --region $region

# Getting the Instance ID from the Server Name
aws ec2 describe-instances --filters "Name=tag:Name,Values=$($ServerName)" --query "Reservations[].Instances[].InstanceId" --profile $chosenProfile --output text --region $region

Download yourself

I've placed the code on my public snippets and you can add this to your profile as well. If you need to create one, you can follow along here. Once you have your profile, just dot source the file into our profile.

# previous profile items
. C:\path\to\Manage-AWSServers.ps1

Find the full code of Manage-AWSServers.ps1 in my PowerShell Snippets

I'm publishing this as part of 100 Days To Offload. You can join in yourself by visiting 100DaysToOffload.com.
Tags: 100DaysToOffload, cloud, powershell, aws

Webmentions

If there are replies, they will show below.

2 Likes



Found an issue? Edit on Github

← Back home