Why your E2E test opened the wrong email

Updated October 16, 2026 · Zyntra

A test reads the wrong email when it reads an address that more than one thing writes to: an earlier run, an earlier step of the same test, or another worker running in parallel. The fix is almost always the same: give each test its own address, and when one address must get two emails, wait for the one you expect by its content, not by its arrival order.

Cause 1: the address was used before

A hard-coded address such as [email protected] still has last night's email when tonight's run starts. If the new email is slow, "get the latest email" returns the old one at once, and the test opens a link that has already been used. It fails with "link expired", which looks like an app bug.

Fix: build the address per test from something unique: a random suffix, or the CI run ID plus a counter.

address.ts
import { randomUUID } from "node:crypto";

// Letters, digits and hyphens only in the tag.
export const newAddress = (tag: string) =>
  `${process.env.ZYNTRA_TEAM_ID}.${tag}-${randomUUID().slice(0, 8)}@zyntramail.com`;

Cause 2: two emails to one address in one test

Sign-up sends a confirmation; the test then asks for a password reset. Both go to the same user, so both go to the same address. Reading "the latest email" right after clicking "Send reset link" returns the confirmation, because the reset is still on its way.

Fix: wait for the email that matches what you expect, usually by subject. Comparing received_at with the time the test clicked works too, but the timestamp comes from the mail server's clock, not the test machine's, so a skew of a few seconds can break it. A subject check has no such problem.

Cause 3: parallel workers share an address

Two workers run the sign-up test against the same address, each reads the other's email, and the failure moves around between runs. Anything that builds an address from the test name alone has this problem as soon as you add a second worker or a browser project. Add the worker index, or use a random suffix as above.

Cause 4: a retry reuses the first attempt's address

If the address is created once per file (in a beforeAll, say), a retried test reuses it and finds the email from the failed attempt. Create the address inside the test, so every attempt starts clean.

A quick checklist

  • Is the address created inside the test, with a unique part?
  • Does the test send more than one email to it? Then match by subject.
  • Does the assertion check the subject before using a link? A wrong email then fails with a clear message.

Questions

Should I delete emails after each test instead?
Unique addresses make that unnecessary, and each delete is another API call. See cleaning up after email tests for when deleting does make sense.
Can a dot separate parts of the tag?
Use hyphens. Zyntra splits the address on dots to find the team and the inbox name, so a dot inside the tag cuts the inbox name short.

Sources