Catch the errors of your apps

Incidents catches the errors and crashes of the apps of your business. An app sends them with any Sentry SDK, and the QA tester looks at each new incident. This page tells how to connect an app, read an incident, and turn one into a bug ticket.

Before you begin

• Incidents is a part of the Engineering product. A person sees it only when it has Engineering.
• Owners and admins add, change, pause and remove the apps.

Connect an app

1. Add the app

   Open Incidents, then the Apps tab. Select Add an app. Give it a name, and say What is it built with?: Web app, Next.js, Node.js, Python or Something else. Select Add the app.

2. Install the Sentry SDK

   The page of the app shows the install command and the code, with its DSN. Put the code Before anything else in your app. For a web app:

   bash

   npm install @sentry/browser

   ts

   1import * as Sentry from "@sentry/browser";2 3Sentry.init({4  dsn: "https://<key>@ingest.troopy.ai/<number>",5  tracesSampleRate: 0,6});

   Any Sentry SDK works. Give it the DSN, and turn tracing off.

3. Send a test error

   Select Send a test error. The app shows Waiting for its first error until the first one comes.

Upload source maps

With source maps, the errors of a release show the own files, lines and functions of your app. Upload the map of each built file of a release from your CI, with the CLI:

1for f in dist/assets/*.js.map; do2  npx troopy incidents sourcemaps upload --id <app id> \3    --release "$RELEASE" --name "${f#dist}" --content "@$f"4done

The page of the app has this script with its id: select Copy the upload script. A map has 25 MB at most. An app keeps the maps of its latest 10 releases.

The settings of an app

The page also says what Troopy dropped today, and why: over a limit, too large, or Over 200 new incidents in a day.

Read the list

Issues lists the incidents of a view:

Each row shows how bad it is, what and where. It also shows the errors of the last 24 hours as bars, the count of errors and of people, and when the last error came. Sort by Last seen, First seen, Most errors or Most people. The keys J and K move between rows. Search incidents finds by title, or by number, such as #14.

Read an incident

The page of an incident shows:

• What happened: the look of the QA tester, with how bad it is, what to do, and why.
• The ticket that fixed it, and the changes that went live just before it began: possible causes.
• The facts, the errors of the last 14 days, and one error whole: The first error, The latest error or an older one.
• The error itself: its frames, What happened before, the request, the person, the device and the tags.

Each word of an error comes from the users of the app. The page shows it as text, and never runs it.

Act on an incident

When the ticket of an incident goes live, Troopy resolves the incident in the next release. The first release after it says "The fix is in release" and its number. An error after the fix brings the incident back, and Troopy notes it on the ticket.

The QA tester and incidents

• Each new incident, each one that came back, and each spike goes to the QA tester for a look.
• A spike is an hour with 100 errors or more, and 10 times the average hour of the day before.
• The QA tester gives each one a severity, a verdict and a summary. It makes a bug ticket only after a person says yes.
• An alert shows in Needs you until a person opens the incident, resolves it or ignores it.

Troopy keeps each error for 30 days.

Related

• Ship code with engineering tickets
• Test your product with the QA tester
• Use the Troopy CLI
• Answer what needs you