DocumentaciónHelp

Manage errors

Review APM's grouped exceptions, mark them as new, known, resolved or ignored, and close their incident with «Declare fix».

On this page

APM groups identical exceptions into a single error: same type, same message (without the data that changes) and same point in the code. Instead of a thousand repeated lines, you see one error with how many times it occurred, since when, in which version and a sample trace.

Before you start

  • View permission in the APM module to review. To change an error's status, Operate permission.

Review errors

  1. Go to APM → Errors.

  2. Choose the Status (All, New, Known, Resolved or Ignored) and the Period.

  3. For each error, review:

    • Error: the exception type, its message and the endpoint where it occurred.

    • Service, In the period and Total.

    • Seen: the last time and since when it has existed.

    • Version: the latest version where it appeared and, if it changed, since which version.

  4. Click Trace to see a real request that failed with that error.

From a service's detail page, Errors and logs tab, the See all button opens this page filtered by that service. To remove the filter, click See all in the notice at the top.

Error statuses

Status

When to use it

New

Just appeared or reappeared. It is the status it arrives with

Known

You have already seen it and someone is working on it

Resolved

You fixed it. It is also set this way when its incident closes. If it appears again, it automatically goes back to New and notifies again

Ignored

It requires no action (for example, an expected validation error)

To change it, choose the status from the list in the Status column.

Errors and incidents

  • A new error opens an incident in Core, except during the first half hour of a newly appeared service (to avoid notifying about what was already failing before it was instrumented).

  • An error that was Resolved and appears again sends the notification An error reappeared and opens a new incident (or is added to the incident if it is still open).

  • For the ARGUNIS service itself, an expired session and an incorrect password or verification code at sign-in do not count as errors (they are already in the audit log).

  • If the service has its alerts muted, no incidents are opened.

APM error incidents do not close as soon as the error stops occurring. They close in four ways:

  1. Declare fix: the recommended way. Open the incident in Core → Incidents, click Declare fix, describe what was fixed or changed and confirm. The incident closes, the description is kept in its history and in the audit log, and the closure is notified as a recovery.

  2. New deployment: if you deploy a new version of the service (with a different service.version) and the error does not appear again in the following 30 minutes, ARGUNIS closes the incident on its own and indicates the version.

  3. Days without recurring: if the error does not appear again within the days set in Close errors that don't recur (days) (2 by default; adjusted in the company's or the service's thresholds), ARGUNIS closes the incident on its own, with the note «resolved: the error did not recur in 2 days», and notifies the closure.

  4. Normal error rate: the «Errors in …» incident (rate above the threshold) closes on its own when the rate has been below the threshold for 30 minutes.

Client errors (HTTP 4xx, such as 404 or 409) stay in APM → Errors and as an event, but do not open an incident.

  1. Resolve manually from the incident.

When the incident closes, in any of these ways, the error changes to Resolved in this list (unless it is Ignored). That way, if it appears again, ARGUNIS detects it as a reappearance and opens a new incident.

It doesn't work the other way around: marking the error as Resolved in APM → Errors does not close its incident. The error's status orders the APM list; the incident is closed from Core.

Related

Search Argunis Monitoring

Type at least 2 characters to search Argunis Monitoring.

↑↓ to navigate · Enter to open · Esc to close