Retries

qavajs can automatically retry failing scenarios a configurable number of times before marking them as failed. This is useful for dealing with intermittent failures caused by network instability or timing issues in the test environment.

Configuration

Set the retry option in your config file:

export default {
    // ...
    retry: 2  // retry each failing scenario up to 2 times
}

Or pass it on the CLI:

npx qavajs run --retry 2

A scenario is retried immediately after it fails. If it passes on any retry attempt it is reported as passing. If it fails on all attempts it is reported as failing.

Retry in CI

It is common to enable retries only in CI where infrastructure-related flakiness is more likely:

// config.ts
export default {
    retry: 0,   // no retries locally
    // ...
}

export const ci = {
    retry: 1,   // one retry in CI
    // ...
}

npx qavajs run --profile ci

Combining Retries with Tags

Mark known flaky tests with a tag and run them with extra retries while keeping the rest of the suite strict:

@flaky
Scenario: Payment confirmation arrives within SLA
  When I complete a payment
  Then I expect 'Confirmation Email' to be received

Run with a higher retry count for flaky tests only:

npx qavajs run --tags "@flaky" --retry 3
npx qavajs run --tags "not @flaky" --retry 0

What to Avoid

Retries mask real failures. Use them as a safety net, not a substitute for fixing unstable tests. If a scenario fails consistently it needs to be investigated, not retried indefinitely.

Good candidates for retries:

• Tests that depend on external services with occasional latency spikes
• Tests that interact with email or async notification delivery
• Tests in shared CI environments with variable resource availability

Poor candidates for retries:

• Tests that fail due to application bugs
• Tests that fail due to incorrect assertions
• Tests that leave dirty state on failure (they may pass on retry because state from the previous run persists)

Cleaning Up After a Failed Attempt

When retries are enabled, ensure your After hooks or fixtures clean up properly so that each retry starts from a known state:

import { After } from '@qavajs/core';

After(async function() {
    // always clear cookies so a retry starts unauthenticated
    await this.playwright.context.clearCookies();
    // always reset cart via API regardless of pass/fail
    await this.playwright.page.request.delete('/api/cart').catch(() => {});
});