I did quickly check the docs, but couldn’t find the answers. Therefore, the question:
Given that one would use latest Sentry SDK for Python and have self-hosted Sentry server, would non responding (down) Sentry server cause performance issues for the application? In other words, does Sentry SDK send HTTP requests in a way that they block the application or is it implemented in a non-blocking way?
FYI. I asked a very similar question after those outages a few days ago caused my application to freeze and timeout. So in PHP at least it does block. Not sure about Python.
No responses yet…
In Python, we definitely use a thread to make requests, so everything is asynchronous by default. You can opt out of this behavior, but you’d have to explicitly do it.
As for PHP, I don’t personally know a ton here, but from what I recall, traditionally in PHP, it’s harder to just spin up a thread to make something async. Also traditionally the process quits when a request is finished, as opposed to being a long running process.
Any language/SDK I can think of besides PHP is async by default.