When deploying to a load balanced elastic beanstalk, an ec2 instance will be "unhealthy" and at some point after the application is deployed, but before the application has fully booted, the load balancer will ping on the /health url (or other as defined) until the application responds.
For part of that time oneagent is seeing these requests, which are failing, and then triggers an issue.
On the one hand, while I am not deploying, knowing if /health is not responding could be useful.
On the other hand, I don't care about the status of /health while I'm in the process of deploying.
What is the best practice for dealing with this? Should /health be excluded from monitoring?
Solved! Go to Solution.
We would not recommend to exclude /health from the monitoring. Instead you could define maintenance window (details in blog) with the scope limited to hosts impacted by the deployment.
You can define maintenance window either manually or using API.