Tag Archives: Stop-EC2Instance

AWS Stop-EC2Instance Needs Complimentary Cmdlet

Well, it’s time to try this again: Can a Tweet get another change to the AWSPowerShell module? Maybe you’ve read a recent post about an inconsistent parameter name between the same parameter, in two corresponding cmdlets. One tweet later and that was on its way to being fixed. You can read about that here: http://tommymaynard.com/twitter-reply-hashtag-aws-tweet-prompts-fix-to-awspowershell-module-2016.

I get that a tweet probably isn’t the preferred method of reporting problems, but it worked before, so why shouldn’t it again? That said, if there is a preferred method, then I’d be happy to use that, although it doesn’t make for as good a story.

So, here’s where we need a fix. The Stop-EC2Instance cmdlet includes a -Terminate switch parameter that indicates it will terminate the instance (think, remove) in addition to stopping it. It works; it really does terminate the instance. The problem here is that the help file indicates it will prompt the user for confirmation unless the -Force parameter is also included. Nope; it doesn’t prompt. Not for me at least. It does terminate the instance, however, and it does so with no questions asked. It didn’t get me, but it might ruin someone’s day someday.

I didn’t try it, but it seems that there’s some built-in protection, although it’s not the default.


So you can see it yourself, here’s two PowerShell commands I ran against two different EC2 instances. In the first command I simply stop an EC2 instance. In the second, I added the -Terminate switch. As I mentioned, it works, but there wasn’t a confirmation prompt. This was tested on AWS Tools for Windows PowerShell and the newest version as of this writing,



My suggestion to resolve this problem is to remove the -Terminate switch from the Stop-EC2Instance cmdlet, and make a Remove-EC2Instance cmdlet. Maybe alias it with Terminate-EC2Instance, as Terminate isn’t an approved verb. In fact, when I first wanted to remove an EC2 instance, I started by hunting for a Remove-EC2Instance cmdlet. Tracking down the -Terminate switch of Stop-EC2Instance was the last place I would’ve looked. It seems to go against all PowerShell conventions. Does Stop-Transcript have an option to remove a transcript as part of its usage? Does Stop-VM in Hyper-V delete the VM? Does Stop-Cluster include a way to delete the cluster. No. Terminating an instance shouldn’t be a part of stopping an instance unless Stop-EC2Instance can be piped to Remove-EC2Instance. Think about the safety that a separate cmdlet would provide (and it’s not like the AWS Tools team is shy about adding new cmdlets).

There’s a couple cmdlets out there that can be used to get the metadata from a compiled PowerShell cmdlet. We use it as a part of writing proxy functions. By using this, I was able to determine the reason why it doesn’t prompt. The problem, as best I can tell, is that Stop-EC2Instance’s ConfirmImpact CmdletBinding argument is set to Medium and the default value of the $ConfirmPreference variable is set to High.


This means that no matter how the Stop-EC2Instance cmdlet is run, it’s never going to prompt for confirmation to terminate, unless someone has modified their $ConfirmPreference value to Medium, and that’s not likely at all. Since a cmdlet, or advanced function, can’t have more than one ConfirmImpact argument, we need the terminate functionality removed from Stop-EC2Instance and the development of a Remove-EC2Instance cmdlet. For comparison, here’s the CmdletBinding attribute for the Remove-ADGroupMembership Active Directory cmdlet. This one always prompts unless we include -Confirm:$false.


I get this fix is much more involved than the last one I brought up (changing the name of a parameter), and would require more work. Additionally, it’s probably one of those things that have to be discussed among the AWS Tools team members and management. Before I close, I should make sure everyone knows that AWS is still new to me, and as such, there may be things that weren’t considered, or known, when this post was written.

Thanks for reading, and enjoy the upcoming weekend.

Hashtag AWS Tweet Prompts Fix to AWSPowerShell Module

I started the Twitter Reply category so I would be able to reply to things I saw on Twitter that needed to be done outside of the 140 character limit. I’ve used it this way a few times, however, today usage is a little different. Although it incorporates Twitter and it’s going to take longer than 140 characters, I’m not actually replying to someone so much. Instead, I am bringing up an event that transpired on Twitter.

First, let me begin by saying how impressed I am with at least one of the developers — I’m guessing he’s a developer — working on the AWSPowerShell module. Last Wednesday, I tweeted that I had started and then stopped an EC2 instance using the Start-EC2Instance and Stop-EC2Instance cmdlets. I noticed a naming difference between identical parameters used by these two, complimentary cmdlets, as I mentioned in my Tweet.

Steve Roberts — a complete stranger to me — replies to my tweet as he must follow the #AWS hashtag. The part the sticks out is that he wrote, “…Fixing…,” as if he was going to correct this difference. Hilarious, right?

Cut to just over 24 hours later and it’s fixed. I downloaded the newest version of the module, installed it, and tested it. It was fixed. The complimentary cmdlets now use the same, -InstanceId parameter.

Did that really just happen? For all I knew, this Steve guy was messing with me, but no, my Tweet really did initiate a (very minor) fix included in the newest version of the AWSPowerShell module. That’s a first. Well, I’ve found a new problem. Maybe I can get that one fixed, too. I’ll be writing about that in an upcoming post and will link it from here, as soon as it’s complete.