UNCLASSIFIED - NO CUI

Skip to content

Update dependency grafana/k6 to v0.46.0

renovate requested to merge renovate/grafana-k6-0.x into development

This MR contains the following updates:

Package Type Update Change
grafana/k6 ironbank-github minor v0.45.1 -> v0.46.0
grafana/k6 minor v0.45.1 -> v0.46.0

⚠ Dependency Lookup Warnings ⚠

Warnings were logged while processing this repo. Please check the logs for more information.

Release Notes

grafana/k6

v0.46.0

Compare Source

k6 v0.46 is here 🎉! This release includes:

  • xk6-browser extension version bump to v1.0.2, which includes breaking changes.
  • Send custom headers to Loki log output
  • Cloud Traces, the new integration between k6 and Tempo
  • Ability to configure TLS per gRPC connection
  • Cloud output v2
  • And many UX improvements, bug fixes, and internal improvements
Breaking Changes
Browser

In this release, the xk6-browser extension version is bumped up to v1.0.2, as it includes multiple breaking changes concerning options, browser lifecycle, and metrics. See the migration guide for making your test scripts compatible with the new version.

Options devtools, env and proxy are deprecated (browser#​868, browser#​870, browser#​872). Additionally, browser launch/connect options are no longer defined in the corresponding JS API methods, instead the test execution related options are now defined inside the browser scenario options (see #​3036), and the other more "environmental options", such as headless, debug, executablePath, are set as ENV vars. Also, the slowMo option is no longer supported, although it might be supported again in the future through a different API (browser#​876).

Metrics also went through a few changes. The Web Vitals metrics are renamed to use the browser_ prefix and short namings (e.g.: webvital_first_input_delay -> browser_web_vital_fid) (browser#​885, browser#​903), and the rating metric is removed, as it is now set as a label in the corresponding Web Vitals metrics (browser#​915). The browser HTTP metrics have also been modified, as these are now split from the HTTP module ones, so there are new browser_ prefixed HTTP metrics, specifically for request duration and failed requests (browser#​916).

Another big change introduced in this version affects the way the browser lifecycle is handled. Users no longer have to explicitly initialize/close the browser instance through the JS API. Instead, a browser instance will be automatically initialized/closed at the beginning/end of each iteration if the browser type is set in scenario options (see #​3036). This also means that the chromium entity from k6/experimental/browser import path is no longer valid. Instead, the browser entity provides access to the browser methods such as browser.newPage() (browser#​910, browser#​944). This change means that the browser.on() method is no longer applicable, and therefore it has been removed (browser#​919).

Additionally, related to import changes, the browser module version is no longer an exported field (browser#​923).

Last but not least, this release also includes constraints on browser contexts usage as now only one browserContext is allowed per iteration, which means that the user can create a new browserContext, close it, and then create it again; but can not have more than one "live" browserContext. Instead, scenarios should be used to separate independent actions in a test (browser#​929, browser#​945).

With all these changes, a simple browser test now looks like:

import { browser } from 'k6/experimental/browser';

export const options = {
  scenarios: {
    ui: {
      executor: 'shared-iterations',
      options: {
        browser: {
          type: 'chromium', // chromium is the only supported browser type so as long as
                            // the option is set, Chromium/Google Chrome will be used
        },
      },
    },
  },
};

export default async function () {
  const page = browser.newPage();
  try {
    await page.goto('https://grafana.com')
  } finally {
    page.close();
  }
}
Deprecations
  • #​3121 deprecates the loadimpact/k6 docker image. We recommend switching to the grafana/k6 image as soon as possible.
New features
Loki log output sending additional headers #​3227

k6 has been able to send logs to Loki for nearly 3 years since v0.28.0 but didn't support any way to authenticate.

Now, it can be configured to send additional headers on every request.

This can be done by using the new header config option, similar to label:

k6 --log-output=loki=http://example.org,header.X-My-Header=123,header.Authorization=mytoken ...

The example above will now send the header X-My-Header with the value 123 and the Authorization header with the value mytoken.

Thanks to @​federicotdn for adding this feature.

Cloud Traces #​3100, #​3202

This release supports the new integration between k6 and Tempo in the cloud. Grafana Cloud k6 and Grafana Cloud Tempo customers will be able to start their traces in k6 using the existing k6/experimental/tracing module to enrich their test run analysis page with metrics and aggregations from tracing data.

The new Cloud traces will work "out of the box" for k6 cloud runs. In case of k6 run execution, the K6_CLOUD_TRACES_ENABLED environment variable has to be set to true.

Ability to configure TLS per gRPC connection #​3159, xk6-grpc#​25

The k6/net/grpc and k6/experimental/grpc modules now support configuring TLS per-connection via ConnectParams. This is useful when connecting to a single gRPC server, using different TLS settings for each VUs, as well as when connecting to multiple gRPC servers from the same VU.

Expand to see an example of the new functionality. 
import grpc from "k6/experimental/grpc";
import { check } from "k6";
import { SharedArray } from "k6/data";
import exec from "k6/execution";

// note: the services in this example don't exist. If you would like
// to run this example, make sure to replace the URLs, and
// the cacerts, cert, key, and password variables.
const grpcArgs = new SharedArray("grpc", () => {
    // Using SharedArray here so that not every VU gets a copy of every certificate a key
    return [
        {
            host: "foo1.grpcbin.test.k6.io:9001",
            plaintext: false,
            params: {
                tls: {
                    cacerts: [open("cacerts0.pem")],
                    cert: open("cert0.pem"),
                    key: open("key0.pem"),
                },
            },
        },
        {
            host: "foo2.grpcbin.test.k6.io:9002",
            params: {
                plaintext: false,
                tls: {
                    cacerts: open("cacerts1.pem"),
                    cert: open("cert1.pem"),
                    key: open("key1.pem"),
                    password: "cert1-passphrase",
                },
            },
        },
    ];
});

const client = new grpc.Client();

export default () => {
    if (__ITER === 0) {
        // Take one config and use it for this one VU
        let grpcArg = grpcArgs[exec.vu.idInTest % grpcArgs.length];
        client.connect(grpcArg.host, grpcArg.params);
    }

    const response = client.invoke("hello.HelloService/SayHello", {
        greeting: "Bert",
    });

    check(response, {
        "status is OK": (r) => r && r.status === grpc.StatusOK,
    });

    console.log(JSON.stringify(response.message));
};

Thanks @​chrismoran-mica for the contribution 🙇‍♂️.

Cloud Output v2 #​3117

After years of great service, we decided to refresh the k6 Cloud output introducing a more efficient end-to-end solution for ingesting the generated tests' metrics. The main change regards the protocol used for flushing metrics that is now a binary-based payload over HTTP.

The new output reduces the resources a load generator uses for tests that produce many metrics. There is no significant difference in the user experience; it's expected to be the same.

The one thing worth highlighting is that the new output is strict about tags, and it'll drops tags if they are reserved. For example:

  • A custom tag named test_run_id, since that is reserved for internal k6 Cloud operations
  • Any tag with a key that starts with two underscores (__), that is marked by Prometheus convention as reserved

This is not yet the default Cloud output for the test runs executed from local machines (k6 run -o cloud), but it is expected to be transparently enabled in the upcoming weeks.

The full list of related MRs: #​3104, #​3108, #​3120, #​3125, #​3162, #​3169, #​3182, #​3186, #​3187, #​3193, #​3195, #​3206, #​3226, #​3157, #​3172.

UX improvements and enhancements
  • #​3176 Adds a js/promises package, which enables extension developers to easily create promises that will be dispatched to the eventloop using the New function.
  • #​3181 Adds a RunOnEventLoop method to the modulestest.Runtime type, which allows extensions developers to run code on the event loop from their tests.
  • xk6-grpc#​19 Fixes EOF (end of file) error that was produced when writing to a closed gRPC stream. Additionally, if no error handlers are set up, the error is logged to the console.
  • browser#​881 Adds a new browser_http_req_failed metric.
  • browser#​901 Adds support for shadow DOM piercing in locator. Thanks to @​tmc for its implementation.
  • browser#​953 Improves Web Vitals reporting for better accuracy and consistency.
Bug fixes
  • #​3178, xk6-grpc#​23 Fixes the registration of nested protobuf messages in the gRPC module. Thanks @​thiagodpf for reporting and fixing the bug!
  • #​3194 Fixes loki log output exiting before push is finished.
  • #​3231 Fixes the tracing module sampling option to default to 1.0 when not set by the user.
  • browser#​866 Disables the timeout on browser process execution.
  • browser#​942 Fixes deadlock in frameNavigated.
  • browser#​943 Fixes a race condition in navigation and lifecycle events.
  • browser#​979 Fixes a deadlock on pagehide event evaluation when page.close() is called.
  • browser#​980 Fixes Loki log output for console log serializer.
  • browser#​991 Fixes data directory not being cleared during browser.close() panic.
Maintenance and internal improvements

Configuration

📅 Schedule: Branch creation - At any time (no schedule defined), Automerge - At any time (no schedule defined).

🚦 Automerge: Enabled.

♻ Rebasing: Whenever MR becomes conflicted, or you tick the rebase/retry checkbox.

🔕 Ignore: Close this MR and you won't be reminded about these updates again.

  • If you want to rebase/retry this MR, check this box

This MR has been generated by Renovate Bot.

Merge request reports

UNCLASSIFIED - NO CUI