InnerWeb
Skip to main content

Case: shared error monitoring for several installations

A vendor that supports many deployments can run one shared monitoring of unhandled exceptions for all installations. The on-duty engineer then sees one list of error types instead of a dozen panels. This case shows how such a setup is organized, how it groups errors, and which checkpoints are worth verifying with the vendor.

1
shared error monitoring for all installations
Source: function documentation, error reporting
5
checkpoints to add to the vendor questionnaire
Source: InnerWeb study

Situation: one monitoring project for many installations

A software vendor that supports several deployments has two options for error monitoring: a separate project for each installation, or one shared project. In the case described, the shared setup was chosen. All installations report unhandled exceptions to a single project, and every event is tagged with the installation and the release version.

The benefit is obvious: the on-duty engineer sees one list and can check whether the same error affects one installation or several. A shared setup does, however, require discipline: data from different plants must not mix inside message content, and only people authorized on the vendor side may access the project.

How the system groups errors

The monitoring groups events into error types, that is, groups with the same cause. As a result, a recurring error is one entry with a counter, not hundreds of separate records. The on-duty engineer sees which error type is new, which one is growing, and which has been present for years.

Grouping works well only when error messages are stable. An error with an identifier glued into its text would split into many seemingly different types and make the scale hard to assess. For this reason, the vendor should strip variables from messages and keep identifiers in separate event fields.

What is excluded from reporting

Exceptions describing invalid input data, such as no access, a missing page or a malformed request, are not system failures, so they are excluded from reporting. This reduces noise and lets the team focus on errors that require a code fix.

Occupational safety alarms are never muted. The mechanisms that tidy up error monitoring must not limit or delay notifications about a threat to life and health: that path is isolated from them.

What the IT on-duty engineer sees

The on-duty engineer sees one project with a list of error types, their counters, the release version and the installation each one concerns. This makes it possible to answer three questions quickly: did the error appear after the last update, does it affect one installation or all of them, and does it touch safety functions.

User reports add context to this picture: what the operator was doing and what effect they saw. The guide on reporting errors describes how to pass such context so that the vendor can reproduce the problem without further questions.

Checkpoints for the plant assessing the vendor

A plant that assesses a vendor can add questions about the organization of error monitoring to its questionnaire. Each question should have an answer that can be verified with a demonstration or a document.

A good vendor answer does not stop at naming a tool. It describes the process: who reviews errors, how quickly they react, and how they inform the client about a fix.

  • Do errors from all installations reach one monitoring, and how is data of individual plants separated?
  • Does every event carry the release version and the installation tag?
  • How does the vendor group errors and clean messages of personal data?
  • Are occupational safety alarms isolated from the mechanisms that tidy up monitoring?
  • Who reviews new error types, and how quickly do they react?

Conclusions for the plant

Shared error monitoring gives the vendor a full picture of deployment quality and shortens the time to detect a problem, while the plant gets a faster fix. The condition is order: stable messages, a tag for version and installation, exclusion of non-failures, and isolation of safety alarms from the rest of the mechanisms.

Ask the vendor about these elements before signing the contract, and after deployment include them in the periodic review of service quality, together with the rules for reporting errors and accepting updates.

Video

Frequently asked questions

Does shared monitoring mix the data of different plants?

It should not. Events are tagged with the installation, messages are cleaned of personal data, and only authorized people on the vendor side have access to the project.

Do user errors, such as no access, reach the monitoring?

No. Exceptions describing no access, a missing page or a malformed request are excluded from reporting as non-failures.

Can monitoring mechanisms mute a safety alarm?

No. Occupational safety alarms are never muted and are isolated from the mechanisms that tidy up error monitoring.

Ask us how we run error monitoring

We will show how the IT on-duty engineer sees system errors and how reports are handled. Book a conversation with the InnerWeb team.

Let's talk