EC2 and other cloud VM inventory status best practices?

[dmathias]

Hi - we run 2019 R1 On-Prem.

In the new account I'm supporting, I'm just now dealing with cloud inventory for the first time.  Azure and AWS EC2 instances are in scope.  Inventory practices have been in place long before I took over the account.

I see stale inventory for cloud VMs in our Active Inventory bucket.  When I look at the Cloud Hosting tab, I see the last state was "Terminated".  Based upon documentation, I would assume that means that instance is not coming back and cannot come back with the same serial or instance ID.  So, that would be an argument for flipping the stale inventory status from Active to Ignored.

However, looking at the History tab for this inventory device, I see it has flipped status repeatedly between Started and Terminated over time.  That leads me to believe that "Terminated" is not actually the true instance state.

How do I look at this apparent contradiction?

Thanks, David

[JohnSorensenDK]

@dmathias 

Did you find the root cause of your findings above? If not, did you reach out to AWS/Azure to check how this could potentially happen?

As always, please feel free to follow How to Contact Flexera Support if you think that Flexera Support can assist in troubleshooting further.

And please update the forum members about your findings if you think that others would benefit from it.

Thanks,

[dmathias]

Hi John - no input from anyone so far at Flexera or from the Community.

I will open a support case, then.  I was hoping this was already a "known issue" and others in the community had worked through it.

Best, David

[darren_haehnel]

For AWS EC2, Terminated means it's not coming back and it's "highly unlikely" to come back with the same serial number.  However, I have found that if the EC2 instance has been built from an "Amazon Linux AMI", there is no dmidecode installed which prevents the agent from collecting a serial number.  Without a serial number it may be that FNMS is confusing multiple instances as one device, thus flipping the status.  Just a hunch.  

Also, we get a report from all AWS accounts with instance details saved as csv/json in an S3 bucket once daily.  We are not currently using the AWS connector but are managing lifecycle of EC2 instances (setting Ignored using a business adapter) by the absence of the instanceID on the report.  My understanding is that Terminated instances only hang around for a couple of hours so there are many that would not show up on a daily report.

[dmathias]

@darren_haehnel , thank you.   The definition you share is what I also saw in Flexera Help, and so my confusion.  Your comment about build instructions may be helpful, but it will probably take us a bit of time still to identify the customer-owner of that process.  In the meantime, I'll share this thread with the NTT DATA support team.  In our case, these instances may be terminated for many days and then History tab shows them coming back - this is not behavior we would expect from the definition.

Best, David