How to opt-out/in of monitor capturing using @sentry/nextjs

[george-gkas]

Hello Sentry team,

I want to ask if it is possible to opt-out/in of Sentry monitoring during the runtime using the @sentry/nextjs library. Right now, we initialize the Sentry instance through sentry.client.config.js; however, if we want to allow users opt out of client-side monitoring, should we call Sentry.close() from within our React component? And if we want to allow users to enable monitoring again, should we call Sentry.init again? Do you know if this setup is valid?

[AbhiPrasad]

Sentry.close() should close the sentry instance. If you look at the implementation you can see internally it sets the internal enabled flag:

sentry-javascript/packages/core/src/baseclient.ts

Lines 261 to 266 in ef6b3c7

	public close(timeout?: number): PromiseLike<boolean> {
	return this.flush(timeout).then(result => {
	this.getOptions().enabled = false;
	return result;
	});
	}

To re-enable Sentry, you shouldn't call init again. Instead you should do the following:

import { getCurrentHub } from '@sentry/nextjs';

const hub = getCurrentHub();
const client = hub.getClient();
if (client) {
  client.getOptions().enabled = true;
}

Which will make the SDK resume instrumenting your app!

[george-gkas]

Thank you very much @AbhiPrasad, for the quick response. 🚀

Just have a couple more questions so I can better grasp the setup.

• If I don't use server logic in my code (eg. /api, getServerSideProps) should I still define sentry.server.config.js?
• Does Sentry automatically distinguish between server and client errors? Can I group client and server errors on my Sentry dashboard, like having separate views. I would love to see a basic example on how to do so. From my understanding @sentry/nextjs will create separate hubs for client and server, is this correct? Probably I can set the scope in my config files like

Sentry.configureScope(function(scope) {
  scope.setTag("execution-env", "client" // "server");
});

[AbhiPrasad]

If I don't use server logic in my code (eg. /api, getServerSideProps) should I still define sentry.server.config.js?

You'll still need the sentry.server.config.js file, but you can leave it empty (not include any Sentry.init call)

Does Sentry automatically distinguish between server and client errors

You can look at the runtime tag that is set:

runtime: browser for client-side

sentry-javascript/packages/nextjs/src/client/index.ts

Line 52 in ef6b3c7

scope.setTag('runtime', 'browser');

runtime: node for node-side

sentry-javascript/packages/nextjs/src/server/index.ts

Line 110 in ef6b3c7

scope.setTag('runtime', 'node');

You can also configure the DSN for each individual Sentry.init in your sentry.server.config.js and sentry.client.config.js to send them to different Sentry projects, which means you no longer filter by tag, but instead by project!

[george-gkas]

thanks @AbhiPrasad