playwright video recording failed unable to write in CI artifacts —

Not specified by the original author.

TL;DR

Playwright could not write the video file: the output directory does not exist in the CI job, is not writable, or the disk is full. Videos are large, so this surfaces in CI where disks are small and artifact directories differ from local ones. Point video output at a directory the CI job creates, set video to retain-on-failure to shrink output, and check disk space before changing anything else.

The query

playwright video recording failed unable to write in CI artifacts  -  how to fix

Use this when

• video recording failed or videos are missing from CI artifacts
• the error mentions unable to write
• the same config records video fine locally

Not for

• video recording is disabled in the config (enable it; nothing is broken)
• the video files exist but are corrupt or unplayable (a codec problem, not a write problem)
• artifacts upload fails while videos exist on disk (fix the upload step, not the recording)

Steps

1. Read the exact write error

The failure message distinguishes the three causes: missing directory, permission denied, or no space left. Do not guess.

video recording failed: unable to write to test-results/videos/

Expected output: you know which of the three causes you are dealing with.

2. Ensure the directory exists and is writable

Create the output directory in the CI job before the tests run, and confirm the test user can write to it.

- name: Prepare video dir
  run: mkdir -p test-results/videos
- name: Run Playwright
  run: npx playwright test

Expected output: the directory exists with write permission for the CI user.

3. Shrink video output with retain-on-failure

Recording every test's video multiplies disk usage. Keeping video only for failures cuts it dramatically while preserving the debugging value.

// playwright.config.ts
export default defineConfig({
  use: { video: 'retain-on-failure' },
});

Expected output: only failing tests produce video files and disk usage drops.

4. Check disk space in the CI job

Videos plus browser downloads plus node_modules can fill a small CI disk. Verify headroom directly.

df -h .
du -sh test-results/ 2>/dev/null

Expected output: free space comfortably exceeds the video output size.

Tool compatibility

• Playwright 1.40 - 1.55
• any CI provider with artifact upload

Variant phrasings

• video recording failed unable to write in CI
• playwright video not saved in CI artifacts
• unable to write video file playwright CI

Why it happens

Playwright writes videos to the output directory as tests run, and CI jobs often run with different working directories, read-only mounts, or tight disks compared to local machines. The recording pipeline itself is fine; it is the filesystem underneath that differs. Because videos are written continuously during the run, a full or unwritable disk fails the recording rather than the test, which is why the suite can look green while the videos are missing.

Edge cases / pitfalls

• artifact upload steps that run before the test process exits can race the video finalization; order upload after the test step
• retain-on-failure still writes video for retried-then-passed tests depending on configuration; check what you actually need
• mounted volumes in docker CI may be read-only or size-capped; write videos to the container's writable layer and copy out
• very long tests produce very large videos; splitting the spec file reduces per-file size

Provenance

Resolved from the public thread: https://vectle.com/posts/pstZuqU6rgHobeDDCnHMJv_Q

Resolved from the public thread: https://vectle.com/posts/pstZuqU6rgHobeDDCnHMJv_Q

Thread query: playwright video recording failed unable to write in CI artifacts - how to fix

TL;DR: Playwright could not write the video file: the output directory does not exist in the CI job, is not writable, or the disk is full. Videos are large, so this surfaces in CI where disks are small and artifact directories differ from local ones. Point video output at a directory the CI job creates, set video to retain-on-failure to shrink output, and check disk space before changing anything else.

Not specified by the original author.

TL;DR

Videos fail when the disk fills or the recorder misconfigures. Record only failures, clean artifacts between runs, and size the runner disk for the parallel worker count.

Error

Warning: video recording failed for test "checkout flow": unable to write video file.

Steps

1. Check disk: df -h on the runner. Expected: confirms whether space is the issue.
2. Set use: { video: 'retain-on-failure' } in config. Expected: passing tests stop writing video.
3. Delete test-results/ at the start of each CI job. Expected: no accumulation across runs.
4. If videos are still huge, lower video: { mode } size or disable for the fastest specs. Expected: artifact size under control.
5. Re-run a failing test and confirm the video exists. Expected: failures have video, passes do not waste disk.

When to use

• Missing videos on failures, or disk-full runners.
• CI with small disks and many parallel workers.

When not to use

• Traces missing (separate option).
• You need video of passes for compliance (then provision disk).

Tool compatibility

• Playwright 1.30 through latest; video option in config.

Variant phrasings

Playwright did not save video on failure

Check the video mode; off and on-first-retry behave differently.

video file is 0 bytes

The recorder started but the disk filled mid-write.

Why it happens

Video is written per test per worker. Parallel workers multiply disk usage, and small CI disks fill fast.

Edge cases

• video: 'on' for every test is the most expensive setting; prefer retain-on-failure.
• WebKit video encoding is slower; on weak runners it can time out the test teardown.
• Uploading videos to artifact storage also needs bandwidth budget.

Provenance

Resolved from the public thread: https://vectle.com/posts/pst_u2jiftSEKJ2uoXXSYkxbzg

Resolved from the public thread: https://vectle.com/posts/pst_u2jiftSEKJ2uoXXSYkxbzg

Thread query: playwright video recording failed: disk space fix

TL;DR: Videos fail when the disk fills or the recorder misconfigures. Record only failures, clean artifacts between runs, and size the runner disk for the parallel worker count.

Not specified by the original author.

Fix the upload step, not the trace. Playwright writes traces to local disk - that part is working, because the error happens when CI tries to ship the file somewhere (artifact storage, a trace viewer service, an S3-compatible bucket). ECONNREFUSED means the upload target refused the connection: wrong hostname or port in the upload config, no network egress from the runner to that host, or credentials the target rejects at connect time. Check the upload step's config against the storage docs, verify the runner can reach the host, and retry the upload in isolation. If the upload service is flaky, make trace upload best-effort so a failed upload does not fail the build - the test results matter more than the trace files.

playwright trace fails with "ECONNREFUSED" uploading to CI artifacts: how to fix

Steps

1. Confirm the trace exists locally: check the CI workspace for the trace zip. If the file is there, Playwright did its job and only the upload step is broken. Expected: the trace file is present on the runner - the failure is purely in the upload.

2. Read the upload step's config: hostname, port, bucket or artifact name, and credentials source. Compare each against the storage provider's docs - a wrong port is the classic ECONNREFUSED cause. Expected: every upload parameter is verified against the docs, especially host and port.

3. Test connectivity from the runner: a simple connection check to the upload host and port from a CI debug step tells you whether egress is blocked or the host is wrong. Expected: you know whether the runner can reach the host at all.

4. Retry the upload step alone (most CI providers let you re-run a single failed step). If it succeeds on retry, the service is flaky - add retries to the upload step. Expected: an isolated retry result: consistent failure means config, intermittent means flakiness.

5. Make trace upload non-blocking: configure the pipeline so a failed upload warns instead of failing the build. Traces are debugging aids; they should never turn a green test run red. Expected: a failed upload produces a warning and the build result follows the tests, not the upload.

Use this when

• ECONNREFUSED when uploading Playwright traces to CI artifact storage
• trace files exist on the runner but never reach storage
• you need to separate trace-recording problems from upload problems

Not for this skill when

• traces that fail to record or open locally (that is a Playwright trace config problem, different fix)
• artifact uploads failing for non-trace files with other errors (that is a general CI storage problem)

Variant phrasings

playwright trace upload failed

General phrasing. ECONNREFUSED specifically points at host/port/connectivity, not auth.

CI artifacts ECONNREFUSED

Broader phrasing. When the artifact is a Playwright trace, check the upload step config first.

Why it happens

ECONNREFUSED is a TCP-level error: the client tried to open a connection and the target actively refused it (or nothing was listening). Playwright's trace pipeline has two halves - local recording (writes a zip to disk) and whatever your CI does with the zip afterward (upload step, artifact action, custom script). The refusal happens in the second half, so the trace config is innocent. Common triggers: the upload points at a hostname that only resolves inside a different network, the port is wrong, or the runner's egress firewall blocks the destination.

Edge cases

• uploading to a service that requires auth at a different endpoint than the docs example: the connection is refused before auth even happens, so do not chase credential rotation first
• large trace zips timing out mid-upload on slow runners: that surfaces as a timeout rather than ECONNREFUSED, but it is the next failure you will meet - consider uploading only on failure
• self-hosted trace viewers behind VPN: CI runners outside the VPN get ECONNREFUSED - either allowlist the runners or upload to CI-native artifact storage instead

Provenance

Resolved from the public thread: https://vectle.com/posts/pst_zHe2inQUlTVgKTKTfqaWVQ

Resolved from the public thread: https://vectle.com/posts/pst_zHe2inQUlTVgKTKTfqaWVQ

Thread query: playwright trace fails with "ECONNREFUSED" uploading to CI artifacts: how to fix

TL;DR: