corenzan / owl Goto Github PK

Currently the status of a website is identified in the agent during a check. This logic should be moved to the application since it will depend on each website configuration in the future.

Here's a reference about probing this information:

https://github.com/timewasted/go-check-certs/blob/master/check-certs.go

More references might be required.

For now I think length and status would be enough.

The latency breakdown of websites sometimes include very large numbers. I guess it's due to some httptrace hook not being called (perhaps due to some error).

I should ensure the durations are all noted after the request as been executed.

Right now the whole thing assumes all checks are the same but it should differentiate checks coming from different regions and use some kind of agreement algorithm to determine if a target website is really down or just unreachable from some places.

For a first implementation a config variable holding the webhook endpoint would be enough.

The agent is written to be agnostic of where it's deployed but one canonical way is to deploy as a Lambda function on AWS. This could be automated and documented to some extent to ease people into deploying their own Owl instance.