Payments are experiencing issues due to temporary restrictions in Russia. If your payment does not go through, please submit a support request.Our support team is available 24/7 — we are always here to help with hosting and server issues.We are now accepting requests for dedicated server rental and colocation services in our data center.Reminder: we recommend enabling backups for additional data protection.A new VPS/VDS lineup with NVMe storage and improved performance is now available.Maintenance work on some servers has been completed. All services are operating normally.
Article3 min read

Site Is Up, Orders Fail: How to Configure Useful Store Failure Alerts

Yandex Monium has updated its alerting system. I examine which signals an online store needs and why checking the homepage does not replace monitoring actual sales.

Duty service phone on the desk in the evening
In this article

Visitors open the homepage, monitoring shows a green status, yet purchasing is impossible because the shipping calculator is unresponsive. Formally, the site is available. For a store, this is still a failure because a critical user journey has stopped.

The September Yandex Cloud digest describes Monium updates: displaying the time in the current alert state and simplified configuration of signals based on error budget consumption. This news is useful as a reason to review the signals themselves. Even a convenient dashboard will not help if it monitors the wrong things.

Identify actions that cannot be lost

For a store, these are typically viewing a product, adding it to the cart, and completing the order. However, the list depends on the business model. For a wholesale catalog, sending a request is critical; for a store with pickup, selecting an available location is key.

Checks must account for dependencies: the database, search, shipping, payment service, and email. Artificial tests must not create real purchases or send messages to customers. Developers should implement a safe control scenario.

It is useful to distinguish between technical availability and business outcomes. A server may respond quickly but return an empty product list. For infrastructure, this is a successful response; for a shopper, it is a broken store.

Alerts must trigger action

A message stating 'high load' without a timestamp, context, or responsible person quickly becomes background noise. Useful notifications should show which operation is affected, how long the issue has lasted, and where to find details.

Thresholds should be set based on normal system behavior. A short spike during scheduled processing may be acceptable, while a slight but prolonged degradation is not. Frequent false alarms teach teams to ignore alerts.

An error budget is a way to predefine the acceptable share of failed operations over a given period. It helps distinguish between isolated failures and rapid quality degradation. However, you must first define what counts as a successful operation from the shopper's perspective.

Verify who receives alerts at night

Settings must include a real recipient, a backup contact, and an escalation procedure. If an alert goes to a general chat without a designated owner, everyone may assume someone else is already handling the issue.

After resolving an outage, it is crucial to re-verify the user journey. Restarting a service does not prove that deliveries and order processing have been restored. I would also track the time elapsed from detection to the return of sales. These metrics demonstrate the value of monitoring far better than the number of charts in a dashboard.

Discussion0

Share your experience and ask questions. Comments without links appear after editorial review.

No comments yet. Start the discussion.