Performance monitoring is still an experimental, work-in-progress feature, and its API may be unstable. It may also have bugs. Feedback is appreciated, and should be submitted as an issue in the repository linked on the right side of this page.
If you’re on a legacy plan, you'll need to add transaction events to your subscription to use performance monitoring.
With performance monitoring, Sentry tracks your software performance, measuring metrics like throughput and latency, and displaying the impact of errors across multiple systems. Sentry captures distributed traces consisting of transactions and spans, which measure individual services and individual operations within those services. Learn more about our model in Distributed Tracing.
If you’re adopting Performance in a high-throughput environment, we recommend testing prior to deployment to ensure that your service’s performance characteristics maintain expectations.
Sampling for transactions must also be configured before tracing is be enabled in your app. Set the sample rate for your transactions by either:
- Setting a uniform sample rate for all transactions using the
traces_sample_rateoption in your SDK config to a number between
1. (For example, to send 20% of transactions, set
- Controlling the sample rate based on the transaction itself and the context in which it's captured, by providing a function to the
The two options are meant to be mutually exclusive. If you set both,
traces_sampler will take precedence.
#include <sentry.h> sentry_options_t *options = sentry_options_new(); // The native SDK currently only supports uniform sample rates. sentry_options_set_traces_sample_rate(options, 0.2); ... sentry_init(options);
Learn more about how the options work in Sampling Transactions.
Test out tracing by starting and finishing a transaction, which you must do so transactions can be sent to Sentry. Learn how in our Custom Instrumentation content.
While you're testing, set
1.0, as that ensures that every transaction will be sent to Sentry.
Once testing is complete, we recommend lowering this value in production by either lowering your
traces_sample_rate value, or switching to using
traces_sampler to dynamically sample and filter your transactions.