To monitor Selenium tests effectively, correlate each test with its Grid request trace, inspect timed spans and error events, and compare the same operations across runs, browsers, and nodes. Selenium Grid has OpenTelemetry tracing enabled by default, but you still need to decide where traces go and how your team will view them. Use test metadata such as se:name to connect a trace to a test, then use logs and the test’s own artifacts to investigate what the telemetry suggests.
What observability tells you about a Selenium run
Observability is not a single dashboard or a verdict that a test is flaky. It is the practice of using a system’s outputs to understand what happened, including cases you did not anticipate. OpenTelemetry describes the three common telemetry signals as traces, metrics, and logs. Selenium Grid’s documentation states: “Observability has three pillars: traces, metrics and logs.” (Selenium Grid observability documentation; see also OpenTelemetry’s observability primer.)
As an Amazon Associate I earn from qualifying purchases.
- Traces show the path of a request through the distributed Grid. A trace is made up of timed spans, which represent operations along that path.
- Logs and span events add detail to those operations. Selenium’s documented event information can include timestamps, trace and span IDs, operation attributes, HTTP status, and exception context.
- Metrics can help show broader patterns, such as load or resource use, when the deployed setup exposes the relevant metrics. Do not assume every Grid release or deployment exposes the same metric set or endpoint.
Trace and span IDs let you correlate related operations; attributes and events add context. An error event may include an exception type, message, and stack trace. That evidence can help distinguish a slow WebDriver operation from a possible network, infrastructure, or client-side issue, but it does not establish root cause by itself. Inspect the test, browser, node, and execution environment as well.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Identify how your tests reach Selenium
Before choosing what to collect, map the execution path. A test may use a local WebDriver, a standalone Selenium Server, a hub-and-node arrangement, or a fully distributed Grid. Containerized deployments add another layer to consider. Grid is designed to distribute test execution across browser and operating-system combinations; the Selenium Grid getting-started guide describes its deployment modes and setup.
#1 Best Overall
This topology determines where a request travels and which component’s evidence you need when diagnosing it. Record the Grid release and deployment mode alongside your monitoring configuration. Selenium options and commands can change between releases, so verify examples against the documentation and tooling for the version you actually run.
Give each session a useful identity
When many sessions run concurrently, a trace is much more useful if you can connect it to the right test and CI job. Selenium Grid supports test metadata such as se:name; the value can appear in the Grid UI or be queried through GraphQL, as described in the Grid setup guide.
Use a stable, recognizable test name and carry related identifiers in your surrounding CI and telemetry systems wherever they support it. Useful context includes:
Rank #2
- Suite and test name.
- Build or CI job identifier and commit.
- Browser name and version.
- Operating system or platform.
- Grid deployment or node identity, if available in your environment.
Keep naming consistent across retries and environments, while making each individual run identifiable. That lets you compare the same test across browsers, Grid nodes, builds, and releases without confusing separate sessions.
Inspect Grid traces and event data
Selenium’s official documentation says Grid is instrumented with OpenTelemetry tracing and tracing is enabled by default. The existence of traces does not automatically mean they are exported to a central backend: choose how your team will consume them, and confirm the deployed Grid’s configuration.
- Confirm trace generation. Check the configuration and logs for the specific Selenium Server/Grid release you deploy. The current Grid observability documentation describes its tracing and event logging behavior.
- Choose a viewing route. Selenium documents console trace output and Jaeger as ways to inspect traces. Console output can be useful while configuring or debugging a smaller setup; a trace viewer is more practical when you need to search and follow requests across a distributed deployment.
- Check verbosity and export configuration. The detail available depends on logging and exporter configuration. Selenium’s 2021 explanation shows a FINE logging example and points readers to the server’s tracing information command for version-specific setup guidance. Do not copy an old command or option without checking it against your installed release: Selenium 4 observability article.
- Verify correlation end to end. Trigger a known test session and confirm that the trace or event data includes enough identifiers to locate its request path and connect it to the named test.
Selenium’s 2021 article describes full-stack tracing for the Java client and server, which can help identify whether time is spent in client-side operations or on the server path. Client-side tracing availability and setup depend on the language client and release; verify the current instructions for your own stack rather than assuming the Java example applies elsewhere.
Triage a failed or slow run
For a failed test
- Find the run using its test name and CI/build context.
- Locate the relevant error event and inspect its exception type, message, stack trace, status, and surrounding spans when those fields are present.
- Follow the trace to see which operations completed, failed, or took unusually long.
- Compare the signature with other runs of the same test, then inspect the relevant browser, node, Grid, and client logs or artifacts.
A repeated exception or an intermittent-looking failure is a lead for investigation, not proof of flakiness. Establish whether the test, browser, network, or infrastructure conditions changed before drawing that conclusion.
Do these 3 things before closing this tab:
1Fix the driver behind crashes, sound loss and screen glitches2Repair Windows errors before they cause bigger problems3Scan for outdated or missing drivers - takes under a minuteFor a slow run
- Find the long-running test’s trace and identify where elapsed time accumulates across its spans.
- Compare the same operation with successful or faster runs, taking browser, version, node, and environment into account.
- Use timestamps and available event context to narrow down whether the delay appears in a client operation, a Grid request, or another part of the execution path.
- Correlate the timing with logs and test artifacts before deciding what to change.
Trace durations show where time was observed, not necessarily why it was spent there. A slow span can direct investigation toward a WebDriver call or infrastructure segment, but it is not by itself proof of the underlying cause.
Choose a trace backend and deployment approach
Selenium documents console output and Jaeger visualization for traces. Choose based on how your team will investigate runs, and verify the exporter, configuration, access controls, and retention behavior for your Grid release and environment.
If you are also deciding whether to operate Grid yourself or use hosted browser execution, assess the operational trade-offs rather than assuming all services expose the same evidence:
| Evaluation area | Questions to answer |
|---|---|
| Operational ownership | Who maintains the Grid, browser images, upgrades, and capacity? |
| Execution coverage | Which browser and operating-system combinations are supported, how are versions controlled, and what parallel session capacity is available? |
| Telemetry access | Can traces, logs, and metrics be exported or queried in the tools your team already uses? |
| Debugging artifacts | Are screenshots, videos, console logs, and session metadata available, and how long are they retained? |
| Integration and cost | What CI integrations, access controls, usage limits, data retention terms, and total costs apply at your expected concurrency? |
Verify current provider capabilities and terms directly before selecting a hosted service; the fact that a service offers browser execution does not establish that it exposes the traces or artifacts your workflow requires.
Free tools Windows power users keep installed
One-click scans. No signup required.
Or skip the browser setup
If you need a screenshot of a page involved in a test, ScreenshotNeo can return a PNG, JPEG, WebP, or PDF through one GET request. It complements Selenium observability; it does not replace Grid traces or diagnose a test failure. The API can remove cookie and consent banners, newsletter popups, and chat widgets before capture, and those steps can be turned off.
For example, this cURL request captures a page as WebP:
curl -G "https://api.screenshotneo.com/v1/shot" -d access_key=YOUR_API_KEY --data-urlencode url=https://stripe.com -o shot.webpSee the ScreenshotNeo API documentation for request options. Bot checks or CAPTCHAs, blank pages, timeouts, failed loads, and cache hits cost nothing; each response reports the page verdict and billing status in the X-Page-Verdict and X-Billed headers. Its MCP server provides take_screenshot, get_page_info, and capture_pdf tools for Claude, Cursor, and other MCP clients. The free plan includes 1,000 shots per month with no card; paid plans start at $5 for 3,000 shots. Sign up for 1,000 free screenshots a month with no card.
Common monitoring problems
- No traces appear in the viewer: Confirm that your Grid version is producing traces and that the configured exporter or console route is sending them where you are looking. Check release-specific setup instructions rather than assuming default tracing also configures a backend.
- Trace output is too sparse: Review the deployed logging verbosity and exporter setup. Selenium’s observability detail can depend on configuration; check the current documentation for your release before changing options.
- You cannot tell which test produced a trace: Add a meaningful session name such as
se:nameand ensure the CI/build identifiers are preserved in your pipeline’s logs or telemetry. - A trace shows a delay but not its cause: Use the span and event as a location clue, then inspect client, Grid, browser, node, and network evidence relevant to that operation.
- You cannot find an expected metric: Confirm which metrics the deployed version and exporter expose. Grid’s trace and event documentation should not be treated as a guarantee that every metric is present in every deployment.
- A failure seems intermittent: Compare multiple runs and their execution context. One trace, or a changing error signature, cannot establish that the test is flaky.
Keep the workflow version-aware
Selenium Grid documentation and commands may evolve. Treat the current Grid observability page as the starting point for the release you run, and use older material such as the 2021 article for concepts and historical examples, not as an unverified configuration recipe. Keep the Grid version, client language and version, exporter, and backend configuration with your monitoring setup so future maintainers can reproduce how traces are generated and viewed.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
