ci: alert Slack when nightly e2e fails

[ilya-bogin-keenable]

Что

Добавляет job notify в .github/workflows/e2e.yml, который пишет в Slack при падении ночного e2e-прогона.

Как работает

• needs: e2e — ждёт всю матрицу (Linux/macOS/Windows; fail-fast: false).
• if: failure() && github.event_name == 'schedule' — алерт только на ночном cron, не на ручном workflow_dispatch (там оператор и так видит результат).
• Шлёт в Slack Incoming Webhook через секрет SLACK_WEBHOOK_URL (канал фиксируется самим webhook). Сообщение: статус + ссылка на конкретный run.
• Payload собирается через jq -n --arg, значения JSON-экранируются.
• Если секрет не задан — job не падает, пишет warning и скипается.

Setup

Секрет SLACK_WEBHOOK_URL уже добавлен в репозиторий. Webhook проверен (Slack ответил ok).

🤖 Generated with Claude Code

[qodo-code-review]

Code Review by Qodo

🐞 Bugs (0) 📘 Rule violations (0) 📎 Requirement gaps (0) 🎨 UX issues (0) 🔗 Cross-repo conflicts (0) 📜 Skill insights (0)

1. No cancel/timeout alert ✓ Resolved 🐞 Bug ☼ Reliability

Description

The new notify job only runs when failure() is true, so scheduled e2e runs that end as cancelled
(including common timeout cancellations) will not post a Slack alert even though the nightly run
didn’t succeed. This can silently miss the primary operational signal the job is meant to provide.

Code

.github/workflows/e2e.yml[143]

+    if: failure() && github.event_name == 'schedule'

Evidence

The workflow sets timeout-minutes: 30 for the e2e job, but the new notify job’s condition only
checks failure(), which does not include cancelled outcomes, so timeouts/cancellations won’t
trigger Slack alerts.

.github/workflows/e2e.yml[53-54]
.github/workflows/e2e.yml[140-144]

Agent prompt

The issue below was found during a code review. Follow the provided context and guidance below and implement a solution

### Issue description
The Slack notification job only runs on `failure()`, so it won't notify on cancelled runs (e.g., timeouts). This can cause missed alerts for nightly e2e runs that did not complete successfully.

### Issue Context
The e2e job has a 30-minute timeout, so timeout-driven cancellations are plausible and should be treated as "nightly did not succeed" for alerting purposes.

### Fix Focus Areas
- .github/workflows/e2e.yml[53-54]
- .github/workflows/e2e.yml[140-144]

### Suggested change
Update the `notify` job condition to run on any non-success result for scheduled runs, e.g.:

```yaml
notify:
 needs: e2e
 if: ${{ always() && github.event_name == 'schedule' && needs.e2e.result != 'success' }}
```

This covers both failures and cancellations while still restricting alerts to the nightly schedule.

ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools

---

[qodo-code-review]

PR Summary by Qodo

CI: Notify Slack on nightly E2E workflow failures
⚙️ Configuration changes ✨ Enhancement 🕐 10-20 Minutes

Walkthroughs

Description

• Add a Slack notification job when the scheduled nightly E2E matrix fails.
• Gate alerts to cron runs only, avoiding noise on manual workflow_dispatch runs.
• Post a run link via Incoming Webhook, safely JSON-encoding payload with jq.

Diagram

graph TD
  A["GitHub Actions: E2E workflow"] --> B["e2e job (matrix)"] -->|"needs; only if failure() & schedule"| C["notify job"] --> D{{"Slack webhook"}}

Loading

High-Level Assessment

The following are alternative approaches to this PR:

1. Use slackapi/slack-github-action

• ➕ Less custom shell scripting; purpose-built Slack payload support
• ➕ Easier to extend to blocks/attachments and richer formatting
• ➖ Adds a third-party action dependency and versioning/supply-chain considerations
• ➖ Still requires secret management; not fundamentally simpler for a single message

2. Create a GitHub Issue / use GitHub Notifications instead of Slack

• ➕ No external webhook/secrets required
• ➕ Keeps alerts and history within GitHub
• ➖ Doesn’t reach the team’s primary chat channel as quickly
• ➖ Can create noisy issue churn unless carefully managed

Recommendation: The current approach (a small notify job using curl + jq with a webhook secret) is a good fit: minimal dependencies, runs only on scheduled failures, and avoids brittle manual JSON string interpolation. Consider adopting slackapi/slack-github-action only if you expect richer message formatting or more complex routing later.

File Changes

Other (1)

e2e.yml Add Slack notify job for scheduled E2E failures +30/-0

Add Slack notify job for scheduled E2E failures

• Introduces a new 'notify' job that depends on the full E2E matrix and runs only when the workflow is triggered by 'schedule' and the matrix fails. The job posts a JSON-escaped payload (repo + run link) to a Slack Incoming Webhook from 'SLACK_WEBHOOK_URL', and self-skips with a warning if the secret is unset.

.github/workflows/e2e.yml

---