Useful signals.
Instead of a flood of notifications, checks focus on processes that matter to the business.
Instead of a flood of notifications, checks focus on processes that matter to the business.
Every significant deviation has a recipient and an action process.
We check that the database and required files allow a business operation to be performed again.
We see more than website availability, including failures in registration, booking or data exchange.
Business action → signal → responsible person → checked action.
Backup and files → isolated restore → data and workflow check.
A green signal or successful copy is not proof of a working customer journey.
A successful copy is not proof of recovery. If the scenario fails, the responsible team investigates and repeats the check before accepting the result.
A server can respond while an enquiry fails to reach the CRM. Storage can run out before the next working day. An archive can exist without containing the files an application needs.
We connect observation to specific consequences: what will stop working, which signal will reveal it and who can decide. We do not collect dozens of metrics merely to fill a dashboard.
We select safe checks for the application, database, file storage and necessary external services. Where possible, we check the user journey without creating a real order, payment or mailing. A page check does not replace checking the whole chain.
We observe selected load, error and latency indicators. Thresholds reflect the work context and the time needed to respond. An alert should help someone act rather than become noise that is routinely ignored.
We establish the data needed for recovery: database, files, a compatible application version and required settings. We choose a suitable way to obtain the agreed set and check access to it. Simply copying directories may not be enough: we choose the backup method to suit the database.
In an agreed isolated environment we restore the selected set and test the required user actions. Real emails, jobs and external calls must be restricted before safe testing. We record the result, time taken and discovered limitations.
Checks must account for both sets and access rights. A home page that opens does not prove a participant can continue learning.
It needs an alert for exchange delays and a clear way to stop unconfirmed availability promises. High website availability does not make stale data current.
Identify a critical scenario. What must a user be able to do, and which systems are involved?
Choose technical and business checks. Availability, database, background jobs, external connections and the required confirmation.
Send the signal to an owner. Who should be notified and under what conditions, and how can a false alarm be recognised?
Verify the return to work. Restore the required data and confirm a real result for the person.
Limit: For backups we separately agree acceptable data loss and recovery time. These targets become a commitment only with a confirmed architecture, tests and agreement, not merely because a provider has a backup feature.
The team receives selected checks, clear notifications and instructions. For recovery, it receives the data set, sequence of actions and a record of an actual test with limitations. Further improvements are chosen by business risk, not only by the number of red indicators.
For Beauty, the path from service page to booking and confirmation matters; for an academy, a learner's access to their own cohort's materials and events; for a supplier, opening the catalogue and passing an order to an owner. HTTP 200 or a running Docker container guarantees none of these outcomes.
When configuring monitoring, pNet defines the signal, its source and the response owner. It separately checks that a backup contains agreed records, files and the required application version, and that recovery supports a control scenario in a safe environment. Recovery targets, retention and alerts depend on the project.
Support helps respond to an exception; DevOps helps retain a known version through changes. A graph without action by an owner remains just a graph.
Note: a backup file does not mean recovery of the PostgreSQL database, Odoo files and agreed user workflow has been tested.
A backup alone does not prove that a system can be restored. Recovery targets, retention, notification coverage and any availability commitments require a separate agreement and a suitable test.
Describe your systems, the problem and any known constraints. You do not need a complete technical specification. Please leave passwords, keys and customer data out of your message.
We first establish their contents, retention, availability and recovery process. Then we check whether they are sufficient for your application and acceptable scenario.
No. It helps detect selected deviations sooner and start an appropriate response. Unknown and external causes do not disappear entirely.
A separate safe environment and agreed scope are planned. Any actions that could affect production or external services require separate permission.
Only after the scope and conditions are defined and tested. A target and an actually verified result must remain distinct.
Tell us what must continue working and which data cannot be lost. We will choose useful checks and a first safe recovery scenario.