How to disable automatic instrumentation within a context? | Sentry Community | Page 1

merry marsh Sep 18, 2025, 12:20 PM

#

I'm using Sentry Python, and have an ETL script that makes a lot of DB calls. I'm in a Sentry trace context, and there's a portion of code where it repeats a bunch of SQLAlchemy SQL queries with pagination, and since there are a lot of pages, I get way too many spans.
Generally I live the automatic instrumentation, but is it possible to disable automatic instrumentation for a portion of a code?

Here is what ChatGPT hallucinated for me as what I can do which does not seem to exist in Sentry, but would be accurate to what I'm looking for. Is there anything like this in Sentry Python?

arctic ridge Sep 18, 2025, 12:24 PM

#

Unfortunately, we do not have a suppress_instrumentation api 🙂 (Maybe JS SDK has something liek this, but Python does not have)

#

You can use before_send_transaction: https://docs.sentry.io/platforms/python/configuration/options/#before_send_transaction

Options | Sentry for Python

Learn more about how the SDK can be configured via options. These are being passed to the init function and therefore set when the SDK is first initialized.

#

This is a hook that is called when a transaction is finished (a transaction is the basically a root_span that has all spans that where added in one call to your etl script in it)

#

In there you get the transaction object and you can delete spans from that transaction.

#

Its not the most comfortable way to do what you want, but it works

merry marsh Sep 18, 2025, 12:40 PM

#

that is a shame.
another alternative is to avoid Sentry span/transaction context except for the bits where I'm sure that I want to log all spans within it

#How to disable automatic instrumentation within a context?