InstanceMonitoring is set to false in newly created launch configs

[t0mpson]

Hi,

Using the latest asgard trunk (mainly in order to get support for the latest instance types) , I see that newly created launch configs are set to enableInstanceMonitoring = false.

Is there a reason why this default changed compared to AWS default ? (and previous asgard versions ) ?

I tried enabling this through Config.groovy but I'm proably doing this wrong as this has no effect :

cloud {
    ...
    launchConfig {
        enableInstanceMonitoring = true
    }
}

Looks like this was changed recently but the commit references ASGARD-1273 for which I do not have access: https://github.com/Netflix/asgard/commit/8afe3b73d8b4abe78909ddeb21f4a065a52649e1 Thanks, Tom

[joesondow]

It was changed because the default of true was causing excess costs.

The costs were associated with cloudwatch - primarily storage costs for detailed instances metrics. These metrics are not highly used and disabling just switches from 1min to 5min reporting so they are not gone all together.

Your config setting looks correct, so it's unclear to me why you're not getting the behavior you want. Maybe there's something else in your Config.groovy that is interfering with this snippet, or maybe your Config.groovy isn't getting loaded for some reason, like if it's in the wrong place on the file system.

The change was made in https://github.com/Netflix/asgard/pull/528 for internal Jira ticket ASGARD-1288.

[t0mpson]

Thanks for the explanation Joe.

I'm pretty sure the config file is read , as when I add log4j options there, they are honored. And the section I added is at the end of the cloud {} section, so I don't think it gets overridden.

Is there a way to enable debug / view the loaded config params and how they are set in runtime?

I have seen 2 things in addition:

1. When using Cluster of ASGs, even when the existing ASG has enableInstanceMonitoring=true , the next ASG sets it to disabled. (according to the code, it should copy the existing setting).
2. From looking at the code, it seems as if the new deployment module is exposing this configuration to the UI, but the usual rolling push / cluster of ASGs do not expose this to the users. Is there a plan to add the same?

[aglover]

@t0mpson I just ran a test by changing our test suite's config service to set instance monitoring to true and that value was propagated. Is it possible your syntax is incorrect?

[njsaunders]

I'm experiencing the same issue with the latest (1.5.1) build of asgard - It worked previously - How can I easily test if this file if being fully processed?

[danveloper]

@njsaunders you can put a println in the config file

println "PROCESSING INTERNAL CONFIG"

[njsaunders]

Thanks Dan - I can't see that appearing in any logs, but I've tried removing my AWS creds from that file and Asgard fails to start at all, which would indicate that it is being read, but that the instance monitoring section is getting overridden?