Configuration

The DSN is the first and most important thing to configure because it tells the SDK where to send events. You can find your project’s DSN in the “Client Keys” section of your “Project Settings” in Sentry. It can be configured in multiple ways. Explanations of the configuration methods are detailed below.

In a properties file on your filesystem or classpath (defaults to sentry.properties):

sentry.properties
Copied
dsn=https://examplePublicKey@o0.ingest.sentry.io/0

Via the Java System Properties (not available on Android):

Copied
java -Dsentry.dsn=https://examplePublicKey@o0.ingest.sentry.io/0 -jar app.jar

Via a System Environment Variable (not available on Android):

Copied
SENTRY_DSN=https://examplePublicKey@o0.ingest.sentry.io/0 java -jar app.jar

In code:

Copied
import io.sentry.Sentry;

Sentry.init("https://examplePublicKey@o0.ingest.sentry.io/0");

There are multiple ways to configure the Java SDK, but all of them take the same options. See below for how to use each configuration method and how the option names might differ between them.

The Java SDK can be configured via a .properties file that is located on the filesystem or in your application’s classpath. By default the SDK will look for a sentry.properties file in the application’s current working directory or in the root of your classpath. In most server side applications the default directory to add resources to your classpath is src/main/resources/, and on Android the default is app/src/main/resources/. You can override the location of the properties file by using either the sentry.properties.file Java System Property or the SENTRY_PROPERTIES_FILE System Environment Variable.

Because this file is often bundled with your application, the values cannot be changed easily once your application has been packaged. For this reason, the properties file is useful for setting defaults or options that you don’t expect to change often. The properties file is the last place checked for each option value, so runtime configuration (described below) will override it if available.

Option names in the property file exactly match the examples given below. For example, to enable sampling, in your properties file:

Copied
sample.rate=0.75

This is the most flexible method for configuring the Sentry client because it can be easily changed based on the environment you run your application in. Neither Java System Properties or System Environment Variables are available for Android applications. Please configure Sentry for Android via code or the properties file.

Two methods are available for runtime configuration, checked in this order: Java System Properties and System Environment Variables.

Java System Property option names are exactly like the examples given below except that they are prefixed with sentry.. For example, to enable sampling:

Copied
java -Dsentry.sample.rate=0.75 -jar app.jar

System Environment Variable option names require that you replace the . with _, capitalize them, and add a SENTRY_ prefix. For example, to enable sampling:

Copied
SENTRY_SAMPLE_RATE=0.75 java -jar app.jar

The DSN itself can also be configured directly in code:

Copied
import io.sentry.Sentry;

Sentry.init("https://examplePublicKey@o0.ingest.sentry.io/0");

Sentry will not be able to do anything with events until this line is run, so this method of configuration is not recommended if you might have errors occur during startup. In addition, by passing a hardcoded DSN you are no longer able to override the DSN at runtime via Java System Properties or System Environment Variables.

The SDK can also be configured by setting querystring parameters on the DSN itself. This is a bit recursive because your DSN itself is an option that you must set somewhere (and not in the DSN!).

Option names in the DSN exactly match the examples given below. For example, to enable sampling if you are setting your DSN via the environment:

Copied
SENTRY_DSN={DSN}/1?sample.rate=0.75 java -jar app.jar

You can, of course, pass this DSN in using the other methods described above.

The following options can all be configured as described above: via a sentry.properties file, via Java System Properties, via System Environment variables, or via the DSN.

To set the application version that will be sent with each event, use the release option:

Copied
release=my-project-name@2.3.12

To set the application distribution that will be sent with each event, use the dist option:

Copied
release=my-project-name@2.3.12
dist=x86

The distribution is only useful (and used) if the release is also set.

To set the application environment that will be sent with each event, use the environment option:

Copied
environment=staging

To set the server name that will be sent with each event, use the servername option:

Copied
servername=host1

To set tags that will be sent with each event, use the tags option with comma separated pairs of keys and values that are joined by a colon:

Copied
tags=tag1:value1,tag2:value2

To set tag names that are extracted from the SLF4J MDC system, use the mdctags option with comma separated key names. This option is only useful when you're using one of the logging integrations.

Copied
mdctags=foo,bar

To set extra data that will be sent with each event (but not as tags), use the extra option with comma separated pairs of keys and values that are joined by a colon:

Copied
extra=key1:value1,key2:value2

Sentry differentiates stack frames that are directly related to your application (“in application”) from stack frames that come from other packages such as the standard library, frameworks, or other dependencies. The difference is visible in the Sentry web interface where only the “in application” frames are displayed by default.

You can configure which package prefixes your application uses with the stacktrace.app.packages option, which takes a comma separated list.

Copied
stacktrace.app.packages=com.mycompany,com.other.name

If you don’t want to use this feature but want to disable the warning, simply set it to an empty string:

Copied
stacktrace.app.packages=

Sentry can use the “in application” system to hide frames in chained exceptions. Usually when a StackTrace is printed, the result looks like this:

Copied
HighLevelException: MidLevelException: LowLevelException
        at Main.a(Main.java:13)
        at Main.main(Main.java:4)
Caused by: MidLevelException: LowLevelException
        at Main.c(Main.java:23)
        at Main.b(Main.java:17)
        at Main.a(Main.java:11)
        ... 1 more
Caused by: LowLevelException
        at Main.e(Main.java:30)
        at Main.d(Main.java:27)
        at Main.c(Main.java:21)
        ... 3 more

Some frames are replaced by the ... N more line as they are the same frames as in the enclosing exception.

A similar behavior is enabled by default in Sentry. To disable it, use the stacktrace.hidecommon option.

Copied
stacktrace.hidecommon=false

Sentry can be configured to sample events with the sample.rate option:

Copied
sample.rate=0.75

This option takes a number from 0.0 to 1.0, representing the percent of events to allow through to server (from 0% to 100%). By default all events will be sent to the Sentry server.

By default, an UncaughtExceptionHandler is configured that will attempt to send exceptions to Sentry. To disable it, use the uncaught.handler.enabled option. Exceptions are sent asynchronously by default, and there is no guarantee they will be sent before the JVM exits. This option is best used in conjunction with the disk buffering system described below.

Copied
uncaught.handler.enabled=false

Sentry can be configured to write events to a specified directory on disk anytime communication with the Sentry server fails with the buffer.dir option. If the directory doesn’t exist, Sentry will attempt to create it on startup and may therefore need write permission on the parent directory. Sentry always requires write permission on the buffer directory itself. This is enabled by default if the AndroidSentryClientFactory is used.

Copied
buffer.dir=sentry-events

The maximum number of events that will be stored on disk defaults to 10, but can also be configured with the option buffer.size:

Copied
buffer.size=100

If a buffer directory is provided, a background thread will periodically attempt to re-send the events that are found on disk. By default it will attempt to send events every 60 seconds. You can change this with the buffer.flushtime option (in milliseconds):

Copied
buffer.flushtime=10000

In order to shutdown the buffer flushing thread gracefully, a ShutdownHook is created. By default, the buffer flushing thread is given 1 second to shutdown gracefully, but this can be adjusted via buffer.shutdowntimeout (represented in milliseconds):

Copied
buffer.shutdowntimeout=5000

The special value -1 can be used to disable the timeout and wait indefinitely for the executor to terminate.

The ShutdownHook could lead to memory leaks in an environment where the life cycle of Sentry doesn’t match the life cycle of the JVM.

An example would be in a JEE environment where the application using Sentry could be deployed and undeployed regularly.

To avoid this behavior, it is possible to disable the graceful shutdown by setting the buffer.gracefulshutdown option:

Copied
buffer.gracefulshutdown=false

In order to avoid performance issues due to a large amount of logs being generated or a slow connection to the Sentry server, an asynchronous connection is set up, using a low priority thread pool to submit events to Sentry.

To disable the async mode, add async=false to your options:

Copied
async=false

In order to shutdown the asynchronous connection gracefully, a ShutdownHook is created. By default, the asynchronous connection is given 1 second to shutdown gracefully, but this can be adjusted via async.shutdowntimeout (represented in milliseconds):

Copied
async.shutdowntimeout=5000

The special value -1 can be used to disable the timeout and wait indefinitely for the executor to terminate.

The ShutdownHook could lead to memory leaks in an environment where the life cycle of Sentry doesn’t match the life cycle of the JVM.

An example would be in a JEE environment where the application using Sentry could be deployed and undeployed regularly.

To avoid this behavior, it is possible to disable the graceful shutdown. This might lead to some log entries being lost if the log application doesn’t shut down the SentryClient instance nicely.

The option to do so is async.gracefulshutdown:

Copied
async.gracefulshutdown=false

The default queue used to store unprocessed events is limited to 50 items. Additional items added once the queue is full are dropped and never sent to the Sentry server. Depending on the environment (if the memory is sparse) it is important to be able to control the size of that queue to avoid memory issues.

It is possible to set a maximum with the option async.queuesize:

Copied
async.queuesize=100

This means that if the connection to the Sentry server is down, only the 100 most recent events will be stored and processed as soon as the server is back up.

The special value -1 can be used to enable an unlimited queue. Beware that network connectivity or Sentry server issues could mean your process will run out of memory.

By default the thread pool used by the async connection contains one thread per processor available to the JVM.

It’s possible to manually set the number of threads (for example if you want only one thread) with the option async.threads:

Copied
async.threads=1

In most cases sending logs to Sentry isn’t as important as an application running smoothly, so the threads have a minimal priority.

It is possible to customise this value to increase the priority of those threads with the option async.priority:

Copied
async.priority=10

By default the content sent to Sentry is compressed before being sent. However, compressing and encoding the data adds a small CPU and memory hit which might not be useful if the connection to Sentry is fast and reliable.

Depending on the limitations of the project (e.g. a mobile application with a limited connection, Sentry hosted on an external network), it can be useful to compress the data beforehand or not.

It’s possible to manually enable/disable the compression with the option compression

Copied
compression=false

By default only the first 1000 characters of a message will be sent to the server. This can be changed with the maxmessagelength option.

Copied
maxmessagelength=1500

A timeout is set to avoid blocking Sentry threads because establishing a connection is taking too long.

It’s possible to manually set the timeout length with timeout (in milliseconds):

Copied
timeout=10000

If your application needs to send outbound requests through an HTTP proxy, you can configure the proxy information via JVM networking properties or as a Sentry option.

For example, using JVM networking properties (affects the entire JVM process),

Copied
java \
  # if you are using the HTTP protocol \
  -Dhttp.proxyHost=proxy.example.com \
  -Dhttp.proxyPort=8080 \
  \
  # if you are using the HTTPS protocol \
  -Dhttps.proxyHost=proxy.example.com \
  -Dhttps.proxyPort=8080 \
  \
  # relevant to both HTTP and HTTPS
  -Dhttp.nonProxyHosts=”localhost|host.example.com” \
  \
  MyApp

See Java Networking and Proxies for more information about the proxy properties.

Alternatively, using Sentry options (only affects the Sentry HTTP client, useful inside shared application containers),

Copied
http.proxy.host=proxy.example.com
http.proxy.port=8080

At times, you may require custom functionality that is not included in the Java SDK already. The most common way to do this is to create your own SentryClientFactory instance as seen in the example below.

Copied
public class MySentryClientFactory extends DefaultSentryClientFactory {
  @Override
  public SentryClient createSentryClient(Dsn dsn) {
    SentryClient sentryClient = new SentryClient(createConnection(dsn), getContextManager(dsn));

    /*
     Create and use the ForwardedAddressResolver, which will use the
     X-FORWARDED-FOR header for the remote address if it exists.
     */
    ForwardedAddressResolver forwardedAddressResolver = new ForwardedAddressResolver();
    sentryClient.addBuilderHelper(new HttpEventBuilderHelper(forwardedAddressResolver));

    sentryClient.addBuilderHelper(new ContextBuilderHelper(sentryClient));
    return configureSentryClient(sentryClient, dsn);
  }
}

To use your custom SentryClientFactory implementation, use the factory option:

Copied
factory=my.company.MySentryClientFactory

Your factory class will need to be available on your classpath with a zero argument constructor or an error will be thrown.

Sometimes linters can warn about types used by the SDK that are not available in your app. This can happen, for example, when using the Android integration due to the base SDK having JNDI lookup that takes no effect in Android.

To get rid of the warning, configure your lint.xml as follows:

Copied
<lint>
    <issue id="InvalidPackage">
        <ignore path="**/sentry*.jar"/>
    </issue>
</lint>

And in your Gradle plugin:

Copied
android {
    lintOptions {
        lintConfig file("path/to/lint.xml")
    }
}

Where path/to/lint.xml is the path to the linting configuration file mentioned above.

Help improve this content
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").