Make cloud provider metadata log more robust

[dloucasfx]

I set the first error to info and the rest to Error.
If we made it through the curl, this means that we are on AWS, so the anything error later on is an accurate error

cc: @jrcamp @keitwb
Signed-off-by: Dani Louca [email protected]

[jrcamp]

Should we change Azure as well for consistency?

[dloucasfx]

updated to remove error from the message and make the integration tests happy

[keitwb]

That didn't seem to make them happy yet.

[dloucasfx]

@keitwb
this is strange, the assertion looks for level=error

signalfx-agent/tests/helpers/assertions.py

Line 186 in 3d1c91b

match = re.search(r"(?<=level=)\w+", line)

And the only error log I have changed is after the initial connection succeeding https://github.com/signalfx/signalfx-agent/pull/1620/files#diff-5c32148c814c31d5bd8804f4de667e16e11040ec1e9190a18373e301a1ab772aR33
so looks like for whatever reasons, the aws metedata server connection is not throwing an error.
Is the CI on AWS?