Massive amounts of 429 errors in the log

[d-Rickyy-b]

Hi there,

when I run my certificate monitoring service for a few minutes, I start receiving massive amounts of http 429 errors. According to #746, a backoff mechanism has been implemented already. What else can I do to avoid 1) massive 429 errors and 2) the spam in my logs.

Cheers,
Rico

[mhutchinson]

There are a number of options you could try. The first is reducing the number of parallel workers. Otherwise, changing code in your version of scanner/fetcher.go to tune your experience opens more options. In runWorker:

• Changing the parameters for the backoff.Backoff; I'd recommend increasing the Min to a larger number, and increasing the Factor should also make the backoffs longer
• Removing the glog.Errorf on line 287, or downgrading it to glog.V(1).Infof so you still have the option of seeing the errors

[d-Rickyy-b]

Hi @mhutchinson and thanks for your reply!

If I understood you correctly, both of these options require me to change the code of this module, right?
I'm using certificate-transparency-go as a dependency and am using the JSONClient. The backoff attribute is private and can't be manipulated from the outside. The code in runWorker is also private and can't be modified.

certificate-transparency-go/jsonclient/client.go

Lines 150 to 157 in 6ecca40

	return &JSONClient{
	uri: strings.TrimRight(uri, "/"),
	httpClient: hc,
	Verifier: verifier,
	logger: logger,
	backoff: &backoff{},
	userAgent: opts.UserAgent,
	}, nil

Is there any way for me as a user to solve (or circumvent) the issue?

[mhutchinson]

The PR I've just merged will downgrade these 429 errors to info messages at verbosity level 2. If you don't override the --v argument that configures glog then you won't see this if you use a build that includes my change.