Implementation Plan

The nightly Windows run on 2026-07-26 (run #30190743916) had two failures:

Failure 1 — Background jobs E2E: "Expected 1 cron job report, but found 2"

File: tasks/e2e-background-jobs/run.mts (around line 520) and tasks/e2e-background-jobs/fixtures.mts (line 79)

Root cause: The cron job is scheduled with cron: '* * * * * *' (every second). By the time confirmJobsRan checks the project directory, the cron job has already fired more than once, leaving multiple report-*.txt files.

Fix options:

Option A (preferred): In confirmJobsRan, record which report files exist before calling jobsWorkoff, then assert that exactly 1 new file appeared after:

// before jobsWorkoff:
const existingReports = new Set(fs.readdirSync(projectPath).filter(f => /^report-.*\.txt$/.test(f)))

// run jobsWorkoff...

// in confirmJobsRan:
const allReports = fs.readdirSync(projectPath).filter(f => /^report-.*\.txt$/.test(f))
const newReports = allReports.filter(f => !existingReports.has(f))
if (newReports.length !== 1) { ... }

Option B (simpler): Change the assertion from !== 1 to < 1 — i.e., require at least one report file rather than exactly one. Less precise but eliminates the race condition entirely.

Option C (also viable): Delete all report-*.txt files at the start of confirmJobsRan before running the assertion. But this could hide bugs where no report is written.

Option A is the most correct: it measures what the test actually cares about (the cron job ran at least once during the workoff).

Failure 2 — Smoke tests: Nx Cloud client bundle not initialized

Log: This usually means multiple copies of the Nx Cloud client bundle are being loaded by Nx

This is an intermittent Nx infrastructure issue (see nrwl/nx#35209). The job failed at the Set up job step before even checking out the code, so it's not caused by any Cedar commit.

Fix: No Cedar code change needed. Options:

Add a retry on the smoke-tests job in .github/workflows/nightly-windows.yml using continue-on-error: true + a retry workflow, or use the nick-fields/retry action for the nx run call
Or bump the Nx version to one that has fixed #35209

Summary of changes needed

File	Change
`tasks/e2e-background-jobs/run.mts`	Pass pre-existing report files into `confirmJobsRan` and diff against them
`tasks/e2e-background-jobs/run.mts` line ~112	Capture report file snapshot before `jobsWorkoff()`
`.github/workflows/nightly-windows.yml` (optional)	Add retry on the Nx Cloud transient failure

Uh oh!

Uh oh!

Nightly Windows CI failure #2146

Description

Activity

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions