Choose which AWS resources are monitored
Discover the virtual machines (EC2 and Lightsail), databases, functions and load balancers in your AWS account and choose what is monitored, where it goes in Infrastructure and how often its metrics a
On this page
Once the AWS connection is created, on the same tab you choose which resources in your account ARGUNIS monitors. It works the same as with Railway: anything you don't choose is not monitored and does not count toward the plan.
ARGUNIS only reads AWS: it never starts, stops or changes anything in your account.
Before you start
An AWS connection. See Connect AWS, Google Cloud, Azure, Cloudflare or Coolify.
Add the monitoring permission to the connection's identity: in What this connection enables, Monitoring row, click Copy policy and paste it as an inline policy for the
argunisuser or role in IAM.Choosing what is monitored is up to operators and administrators.
What is monitored for each resource
Resource | Goes in | Status | Metrics |
|---|---|---|---|
Virtual machine (EC2) | Servers | Running or stopped, and the AWS checks (system and instance) | CPU, network and CPU credits |
Lightsail server | Servers | Running or stopped, and the Lightsail status check | CPU and network (AWS does not report memory or disk: for that, use the ARGUNIS agent on the server) |
Database (RDS) | Databases | Available or not | CPU, connections, free and used storage, free memory, read and write latency |
Function (Lambda) | Other hosts | — | Invocations, errors, p95 duration and throttles |
Load balancer (ALB/NLB) | Other hosts | No healthy targets | Requests, 5xx errors, response time, and healthy and unhealthy targets |
The place is a suggestion: you can choose another one.
Each connection sees a single AWS account. With several accounts (for example, a landing zone), create one connection per account: each resource keeps its account and its inventory group.
Choose the resources
Go to Settings → Monitoring → Cloud services and open the AWS tab.
In AWS resources, if you have more than one connection, choose which one (you can also choose it in the Connection column of the Monitoring row, in What this connection enables).
In Regions, enter the regions separated by commas (for example
sa-east-1, us-east-1; up to 10) and click Query.The resources appear grouped by region. In each row:
Check Monitor.
Choose the Place in Infrastructure and the Inventory group.
Choose the Interval: every 1, 5 or 15 minutes. Below it you see the estimated monthly CloudWatch cost for that resource.
At the bottom you see how many changes there are and the estimated CloudWatch cost of everything chosen.
Click Save selection.
Choose all and None check or uncheck the resources in a region.
AWS charges your account for reading metrics: USD 0.01 per 1,000 metrics. At 5 minutes, a virtual machine costs about USD 0.35 per month. A 1-minute interval costs five times as much.
Virtual machines and databases count as servers in your plan; functions and load balancers, as services. If the plan is not enough, the selection is not saved and you are notified.
Stop monitoring
Uncheck the resource and click Save selection. Its history is deleted.
If a monitored resource no longer shows up in AWS (you deleted it or the identity lost the permission), a notice says so, with the Stop monitoring them button.
While the connection is monitoring resources, it cannot be removed.
What happens next
Each resource appears in Infrastructure, in its place, in the Cloud resources section. See View cloud services.
A stopped resource, one with failed checks or one with no healthy targets opens a critical incident. CPU, storage or errors above the threshold open a notification. Both close on their own when things return to normal.
The resource is one more item in the inventory: CMDB, dashboards and relationships.