# Triage an issue

Triage is how you work through errors: mark what's handled, silence what's
noise, and give an owner to what needs one. Triage state
is shared across your whole team and every Grafana replica — anyone with plugin
access can triage, and every change is attributed and audited.

## Open the issue

On your service's **Issues** tab, click a row to open the issue drawer. The
triage bar sits at the top, above the stack trace and impact.

![Issue drawer — the triage bar (Resolve / Ignore / Assign) above the stack trace](/assets/nais-apm-service-navno-issues-drawer.png)

## Resolve, ignore, or assign

| Action | What it does |
| --- | --- |
| **Resolve** | Marks the issue as handled. It leaves the default (unresolved) view. If it happens again after a later deploy, it comes back as **Regressed**. |
| **Ignore** | Silences the issue as known noise. It leaves the default view and stays out until you unresolve it. |
| **Assign** | Gives the issue an owner. Assigning is independent of status — assigning never resolves or reopens an issue. |
| **Unresolve** | Reopens a resolved or ignored issue back into the default view. |

Status (resolve / ignore / unresolve) and assignment fold independently:
changing the assignee never changes the status, and vice versa.

The default view hides resolved and ignored issues, so the list stays focused on
what still needs a decision.

## Understand "Regressed"

An issue you resolved that **starts happening again after a newer deploy** is
flagged **Regressed** and bubbles to the top of the list. Concretely: its state
is `resolved`, it still has occurrences in the current window, and the resolve
predates the newest deploy marker.

This is a deliberate approximation. When you're looking at a **historical** time
range, occurrences can predate your resolve and still read as regressed —
narrow the time picker to "now" for the live picture.

## Mute an issue just for yourself

A **mute** is personal: it hides an issue from *your* view without touching the
shared triage state your teammates see. Use it to declutter your own list
without deciding anything on the team's behalf. Mutes live in your Grafana user
storage and never affect shared state.

## Who can triage, and is it recorded?

Anyone with access to the plugin can triage, but every action records the
signed-in Grafana user as the actor. The issue drawer
shows the current state; the full ordered history (who did what, when) is kept
as an audit log. Concurrent conflicting actions are last-write-wins per event,
which is benign: two people resolving the same issue both converge on resolved.

> **How this is stored**
> Triage state is an append-only event log in Grafana's own organization
> annotations — shared across replicas with zero extra infrastructure. See
> [How Nais APM works](/observability/apm/explanations/how-nais-apm-works.md) for the details.

## Related

- [Issues and fingerprinting](/observability/apm/reference/issues-model.md) — what makes two
  errors the same issue.
- [Create alerts from templates](/observability/apm/how-to/create-alerts.md) — get told about the next
  spike.