A schedule that fails at 06:00 and tells no one is worse than no schedule: everybody assumes the data is fresh. This level connects Odara to an email server once, then subscribes you to the schedule from 7.2 — on failure, and on something failures don't catch.
Goal
A tested SMTP configuration, and an alert subscription on your schedule that emails you when a run fails.
You'll need
- The schedule from level 7.2.
- An SMTP server Odara can send through. For learning, use a testing inbox: services such as Mailtrap give you free SMTP credentials whose messages land in a web inbox instead of anyone's real mailbox. Your company's mail relay, or a mail provider's SMTP with an app password, work the same way.
- About 20 minutes.
From the provider you need: host, port, security (STARTTLS on 587, TLS/SSL on 465), a username and password, and an address to send from.
Steps
1. Configure SMTP
On the Schedule page, click the envelope button next to + New Schedule. Fill in SMTP Settings:

| Field | Example |
|---|---|
| Host | sandbox.smtp.mailtrap.io (or your provider's) |
| Port | 2525 / 587 / 465 |
| Security | STARTTLS (port 587) or TLS / SSL (port 465); None (plaintext) only for a trusted local relay |
| Username, Password | from the provider |
| From email / From name | [email protected] / Odara |
Type your address under Send test email to and click Send test. Check the inbox: a message titled Odara Community — SMTP test should be there within seconds. Then Save. The password is stored encrypted and shown as dots from now on.
SMTP is configured once per Odara installation, by an admin. Everything after this is per schedule.
2. Subscribe to the schedule
In All schedules, click the bell in the Alerts column of your schedule. Alert Subscriptions opens:

Under Add user, pick yourself and Failures only, then Add. The modes:
| Mode | Emails when |
|---|---|
| Every run | always |
| Failures only | the run fails |
| Successes only | the run succeeds |
| Custom thresholds… | the run took at least slow ms, or wrote fewer than min rows |
Add all users of a role subscribes everyone currently in that role — people who join the role later are not added automatically.
3. Make it fail, on purpose
Temporarily point the schedule's pipeline at a broken input — or subscribe to a schedule on
w5-4-flaky with Stop, which already fails — and press ▶ Run now.

The email's subject is [FAILED] <pipeline>. The body has the trigger, start and finish times,
duration, rows written and — most useful — the error message, the same one Monitor shows under the
name. The dialog's Recent alerts list records every email sent, with SENT or an error if the mail
server refused it.
4. The failure that isn't one
Add a second subscription on yourself — Custom thresholds… with min rows 1. Now a run that
succeeds but writes nothing also emails you, with a [SUCCESS] subject and Rows written: 0 in the
body. That's the alarm for the quiet failures of World 5: a green run with an empty result.
Expected result
- Send test delivers a test email.
- A failing run of the subscribed schedule sends
[FAILED] …with the error in the body. - With a min rows threshold, a run that writes nothing sends an email even though it succeeded.
Common pitfalls
Send test fails with an authentication error. Wrong username or password — or a provider that wants an app-specific password instead of your normal one.
Send test hangs, then times out. Port and security don't match (STARTTLS on 465, TLS / SSL on 587), or the network blocks outgoing mail ports. Try the provider's alternative port.
Setting min rows from the input size. Rows written counts what the targets wrote — for a loop, everything its rounds wrote (288 sales in 5.2). A filter that keeps a tenth of the input writes a tenth of the rows; set the threshold from what a good run reports, not from the size of the source. A failed run reports what it wrote before it stopped.
Nobody got the email, and Recent alerts is empty. No one is subscribed with a mode that matches the outcome — Failures only on a run that succeeded, for example. Subscriptions are per person, per schedule.
What you built
The difference between a scheduled pipeline and an operated one: it tells you when it needs you. The last level of this world uses both alerts and Monitor to fix a pipeline you didn't write.
Check yourself
One question, drawn from a larger bank. Get it right to pass.
Who configures SMTP, and how often?