Comment 1 for bug 1976289

This looks like an api rate limiting situation, where the controller was rejecting too many concurrent api requests. What looks to be interesting is that the code which opens the api connection does not appear to account for the "try again" error and instead of backing off and retrying transparently, errors out. Now it should eventually come good - the error will cause the agent to restart but doing it this way is noisy, as opposed to silently retrying because the error is recognised as "try again".