Troubleshooting
If you need help solving issues with your Sentry JavaScript SDK integration, you can read the edge cases documented below.
The SDK is not sending any data
If you set up the Sentry SDK and it's not sending any data to Sentry:
Check that you have configured a DSN and that you are passing it to the
dsn
option inSentry.init()
.If you are using environment variables to pass the DSN, make sure the environment variables are set in all relevant environments. Additionally, if you are using environment variables inside a framework, check that the framework will include the environment variables in your bundle. Often this means you will have to prefix your environment variables with special prefixes defined by your framework (
NEXT_PUBLIC_
in Next.js,NUXT_
in Nuxt,VITE_
for Vite projects,REACT_APP_
for Create React App, ...).Check that you have disabled any ad-blockers.
Set
debug: true
in theSentry.init()
options and observe your console output when you start your application. The SDK may tell you why it is not sending any data.Check the Stats and Suscription pages in Sentry. You may have ran out of quota.
Updating to a new Sentry SDK version
If you update your Sentry SDK to a new major version, you might encounter breaking changes that need some adaption on your end. Check out our migration guide to learn everything you need to know to get up and running again with the latest Sentry features.
Debugging additional data
You can view the JSON payload of an event to see how Sentry stores additional data in the event. The shape of the data may not exactly match the description.
For more details, see the full documentation on Event Payload.
Max JSON payload size
maxValueLength
has a default value of 250, but you can adjust this value according to your needs if your messages are longer. Please note that not every single value is affected by this option.
CORS attributes and headers
To gain visibility into a JavaScript exception thrown from scripts originating from different origins, do two things:
- Add a crossorigin=”anonymous” script attribute
<script src="http://another-domain.com/app.js" crossorigin="anonymous"></script>
The script attribute tells the browser to fetch the target file “anonymously.” Potentially user-identifying information like cookies or HTTP credentials won’t be transmitted by the browser to the server when requesting this file.
- Add a Cross-Origin HTTP header
Access-Control-Allow-Origin: *
Cross-Origin Resource Sharing (CORS) is a set of APIs (mostly HTTP headers) that dictate how files ought to be downloaded and served across origins.
By setting Access-Control-Allow-Origin: *
, the server is indicating to browsers that any origin can fetch this file. Alternatively, you can restrict it to a known origin you control:
Access-Control-Allow-Origin: https://www.example.com
Most community CDNs properly set an Access-Control-Allow-Origin
header.
$ curl --head https://ajax.googleapis.com/ajax/libs/jquery/2.2.0/jquery.js | \
grep -i "access-control-allow-origin"
Access-Control-Allow-Origin: *
Unexpected OPTIONS request
If your application started to misbehave because of performing additional OPTIONS requests, it is most likely an issue with unwanted sentry-trace
request headers, which can happen when you are using too generic a configuration for our Tracing Integration in the Browser SDK.
To fix this, change the tracePropagationTargets
option during SDK initialization. For more details, see Automatic Instrumentation in our Tracing documentation.
`instrument.js` Line Numbers for Console Log statements
If instrument.js
displays in your console while debugging, add Sentry to your Framework Ignore List by adding this pattern: /@sentry/
Chrome then ignores the SDK stack frames when debugging.
Dealing with Ad-Blockers
When you are using our CDN, ad-blocking or script-blocking extensions may prevent our SDK from being fetched and initialized properly. Because of this, any call to the SDKs API will fail and may cause your application to behave unexpectedly.
Additionally, even when the SDK is downloaded and initialized correctly, Sentry endpoints that need to receive captured data may be blocked as well. This prevents any error reports, sessions health, or performance data from being delivered, making it effectively unavailable in sentry.io.
Furthermore, some browsers, like Brave, have built-in ad-blockers that may block requests sent to our endpoint. Even if users deactivate your domain from blocking, Brave might continue to block requests made from service workers.
You can work around our CDN bundles being blocked by using our NPM packages and bundling our SDK with your app. However, the endpoint blockage can only be resolved by using a tunnel as explained below.
Using the `tunnel` Option
A tunnel is an HTTP endpoint that acts as a proxy between Sentry and your application. Because you control this server, there is no risk of any requests sent to it being blocked. When the endpoint lives under the same origin (although it does not have to in order for the tunnel to work), the browser will not treat any requests to the endpoint as a third-party request. As a result, these requests will have different security measures applied which, by default, don't trigger ad-blockers. A quick summary of the flow can be found below.
Starting with version 6.7.0
of the JavaScript SDK, you can use the tunnel
option to tell the SDK to deliver events to the configured URL, instead of using the DSN. This allows the SDK to remove sentry_key
from the query parameters, which is one of the main reasons ad-blockers prevent sending events in the first place. This option also stops the SDK from sending preflight requests, which was one of the requirements that necessitated sending the sentry_key
in the query parameters.
To enable the tunnel
option, provide either a relative or an absolute URL in your Sentry.init
call. When you use a relative URL, it's relative to the current origin, and this is the form that we recommend. Using a relative URL will not trigger a preflight CORS request, so no events will be blocked, because the ad-blocker will not treat these events as third-party requests.
Sentry.init({
dsn: "https://examplePublicKey@o0.ingest.sentry.io/0",
tunnel: "/tunnel",
});
Once configured, all events will be sent to the /tunnel
endpoint. This solution, however, requires an additional configuration on the server, as the events now need to be parsed and redirected to Sentry. Here's an example for your server component:
// This functionality is now built-in to the Sentry.AspNetCore package.
// See https://docs.sentry.io/platforms/dotnet/guides/aspnetcore/#tunnel
// docs for more information.
// This example shows how you could implement it yourself:
using System;
using System.Collections.Generic;
using System.IO;
using System.Net.Http;
using System.Text.Json;
using Microsoft.AspNetCore;
using Microsoft.AspNetCore.Builder;
using Microsoft.AspNetCore.Hosting;
using Microsoft.AspNetCore.Http;
// Change host appropriately if you run your own Sentry instance.
const string host = "sentry.io";
// Set knownProjectIds to a list with your Sentry project IDs which you
// want to accept through this proxy.
var knownProjectIds = new HashSet<string>() { };
var client = new HttpClient();
WebHost.CreateDefaultBuilder(args).Configure(a =>
a.Run(async context =>
{
context.Request.EnableBuffering();
using var reader = new StreamReader(context.Request.Body);
var header = await reader.ReadLineAsync();
var headerJson = JsonSerializer.Deserialize<Dictionary<string, object>>(header);
if (headerJson.TryGetValue("dsn", out var dsnString)
&& Uri.TryCreate(dsnString.ToString(), UriKind.Absolute, out var dsn))
{
var projectId = dsn.AbsolutePath.Trim('/');
if (knownProjectIds.Contains(projectId) && string.Equals(dsn.Host, host, StringComparison.OrdinalIgnoreCase)) {
context.Request.Body.Position = 0;
await client.PostAsync($"https://{dsn.Host}/api/{projectId}/envelope/",
new StreamContent(context.Request.Body));
}
}
})).Build().Run();
If your use-case is related to the SDK package itself being blocked, any of the following solutions will help you resolve this issue.
The best way to deal with script-blocking extensions is to use the SDK package directly through the npm
and bundle it with your application. This way, you can be sure that the code will be always be there as you expect it to be.
The second way is to download the SDK from our CDN and host it yourself. This way, the SDK will still be separated from the rest of your code, but you'll be certain that it won't be blocked since its origin will be the same as the origin of your website.
You can easily fetch it using curl
or any other similar tool:
curl https://browser.sentry-cdn.com/5.20.1/bundle.min.js -o sentry.browser.5.20.1.min.js -s
The last option is to use Proxy
guard, which will make sure that your code won't break, even when you call our SDK, which was blocked. Proxy
is supported by all browser except Internet Explorer, though there are no extensions in this browser. Also if Proxy
is not in any of your user's browser, it will be silently skipped, so you don't have to worry about it breaking anything.
Place this snippet immediately above the <Expandable>
tag containing our CDN bundle. The snippet in a readable format presents like this:
if ("Proxy" in window) {
var handler = {
get: function (_, key) {
return new Proxy(function (cb) {
if (key === "flush" || key === "close") return Promise.resolve();
if (typeof cb === "function") return cb(window.Sentry);
return window.Sentry;
}, handler);
},
};
window.Sentry = new Proxy({}, handler);
}
If you would like to copy and paste the snippet directly, here it is minified:
<script>
if ("Proxy" in window) {
var n = {
get: function (o, e) {
return new Proxy(function (n) {
return "flush" === e || "close" === e
? Promise.resolve()
: "function" == typeof n
? n(window.Sentry)
: window.Sentry;
}, n);
},
};
window.Sentry = new Proxy({}, n);
}
</script>
Third party promise libraries
When you include and configure Sentry, our JavaScript SDK automatically attaches global handlers to capture uncaught exceptions and unhandled promise rejections. You can disable this default behavior by changing the onunhandledrejection
option to false
in your GlobalHandlers integration and manually hook into each event handler, then call Sentry.captureException
or Sentry.captureMessage
directly.
You may also need to manage your configuration if you are using a third-party library to implement promises. Also, remember that browsers often implement security measures that can block error reporting when serving script files from different origins.
Events with 'Non-Error Exception'
If you’re seeing errors with the message “Non-Error exception (or promise rejection) captured with keys: x, y, z.”, this happens when you a) call Sentry.captureException()
with a plain object, b) throw a plain object, or c) reject a promise with a plain object.
You can see the contents of the non-Error object in question in the __serialized__
entry of the “Additional Data” section.
To get better insight into these error events, we recommend finding where the plain object is being passed or thrown to Sentry based on the contents of the __serialized__
data, and then turning the plain object into an Error object.
Capturing resource 404s
By default, Sentry does not capture errors when a resource (like an image or a css file) fails to load. If you would like it to do so, you can use the following code. (Note: We recommend loading Sentry as early as possible in any case, but this method in particular will only work if Sentry is loaded before other resources.)
document.body.addEventListener(
"error",
(event) => {
if (!event.target) return;
if (event.target.tagName === "IMG") {
Sentry.captureException(
new Error(`Failed to load image: ${event.target.src}`),
);
} else if (event.target.tagName === "LINK") {
Sentry.captureException(
new Error(`Failed to load css: ${event.target.href}`),
);
}
},
true, // useCapture - necessary for resource loading errors
);
Remember to pass in true
as the second parameter to addEventListener()
. Without it, the event handler won't be called, since it's being added to the event target's ancestor rather than the event target itself, and unlike JavaScript runtime errors, error
events resulting from load failures don't bubble, and therefore must be captured during the capture
phase. For more information, see the W3C spec.
Build errors with vite
If you're using the Vite Bundler and a Sentry NPM package, and you see the following error:
Error: Could not resolve './{}.js' from node_modules/@sentry/utils/esm/index.js
This might be because the define
option in your Vite config is string-replacing some variable used by the Sentry SDK, like global
, which causes build errors. Vite recommends using define
for CONSTANTS only, and not putting process
or global
into the options. To fix this build error, remove or update your define
option, as shown below:
vite.config.ts
export default defineConfig({
build: {
sourcemap: true
},
- define: {
- global: {}
- },
plugins: [react()]
})
Missing module `diagnostics_channel`
If you're getting build errors when using any of the JavaScript SDKs mentioning something along the lines of "Cannot find module diagnostics_channel", try to configure your build tool to either externalize or ignore the diagnostics_channel
module. The Sentry Node.js SDK uses this built-in module to instrument Node.js fetch requests. Some older Node.js versions do not have the diagnostics_channel
API, which may lead to build errors when attempting to bundle your code.
Most bundlers have an option to externalize specific modules (like diagnostics_channel
):
Terser plugin build errors
If you're using a custom webpack plugin that utilizes Terser or another minifier, you might encounter build errors due to a known webpack bug. This issue has been fixed in webpack version 5.85.1 and later.
For more details about this issue, see webpack/terser-plugin build errors.
Failed to register 'import-in-the-middle' during build time
After adding sentry.server.config.ts
and building the project, you might get an error like this: Failed to register ESM hook import-in-the-middle/hook.mjs
. You can add an override (npm/pnpm) or a resolution (yarn) for @vercel/nft
to fix this. This will add the hook.mjs
file to your build output. See the underlying issue in the UnJS Nitro project.
Nitro updated @vercel/nft
in Nitro version 2.10.0
, so you might not get this error anymore, and you don't need to add this override/resolution.
package.json
"overrides": {
"@vercel/nft": "^0.27.4"
}
pnpm dev: Cannot find package 'import-in-the-middle'
Sentry injects import "import-in-the-middle/hook.mjs"
in your server entry. This import acts as a hint for node bundlers to really include this file. As pnpm implements a strict dependency isolation, this import might cause problems. Per default, shamefully-hoist
is false
(pnpm docs here) and this prevents accessing non-declared dependencies. You probably don't want to change this setting, so you have to explicitly add the dependency import-in-the-middle
:
pnpm add import-in-the-middle
Nuxt: Server-side Nitro is not sending events
Nuxt builds the server-side Nitro application as ES Modules, which introduces limitations for server-side tracing during development. Currently, trace collection is only supported when building and running the application. Development mode (nuxt dev
) is currently not supported.
When running the build output, Sentry needs to be initialized before running the rest of the application. This is done automatically, but might not work for your use case. Read more about this in installation methods.
We are currently investigating an issue where the server-side is not correctly initialized with a recent update of Nitro (the server-side toolkit in Nuxt).
We are working on figuring this out (see issue here). For the time being, you can add the following overrides to your application:
package.json
"overrides": {
"nitropack": "~2.9.7",
"@vercel/nft": "^0.27.4"
}
If you need additional help, you can ask on GitHub. Customers on a paid plan may also contact support.
Our documentation is open source and available on GitHub. Your contributions are welcome, whether fixing a typo (drat!) or suggesting an update ("yeah, this would be better").