How to filter a DataDog visualization by error status?

Question

I'm sending a data dog trace using gopkg.in/DataDog/dd-trace-go.v1/ddtrace/tracer like so:

func doSomeStuff(ctx context.Context) error {
    span, ctx := tracer.StartSpanFromContext(ctx, "do_some_stuff")
    err := theActualOperation()
    span.Finish(tracer.WithError(err))
    return err
}

In DataDog I can see successful (when err = nil) and failed operations e.g. successful ones

{
    account_id: a_happy_caller
    duration: ...
    language: go
    process_id: 1
}

and failed ones

{
    account_id: an_unhappy_caller
    duration: ...
    error: {
        fingerprint: ...
        message: rpc error: code = Internal desc = ...
    }
    issue: {
        age: ...
        first_seen: ...
        first_seen_version: ...
        id: ...
    }
    language: go
    process_id: 1
}

I can successfully search for traces that failed using a query like "do_some_stuff with an error status":

operation_name:do_some_stuff AND status:error

or alternatively "do_some_stuff with some value for error.fingerprint":

operation_name:do_some_stuff AND @error.fingerprint:*

However this doesn't seem to work in visualizations on dashboards where I want to get the latency of successful runs separately from any measures on failed runs (i.e., our failed runs are maximally distorting the measure for "normal" runs). To this end I tried to place a simple "Query Number" visualization for failed runs with

p50:trace.do_some_stuff{status:error}

which doesn't work. And alternatively

p50:trace.do_some_stuff{@error.fingerprint:*}

which results in a query error.

I've tried all kinds of selectors and brooded for hours over the advanced filtering reference but don't seem to be able to filter on either the presence (or content) of the error field of the trace, nor the status that I would have expected to be able to filter by, nor am I able to find an example of how to approach this measurement scenario - generally having measures on filtered traces.

How can I achieve measuring the latency using only the traces of successful or only failed runs? Or must I send different messages to achieve this?

How to filter a DataDog visualization by error status?

0 Answers0