Source code is the ultimate documentation
So I've been coding a custom Sentry SDK. While developing a Sentry SDK they recommend you to run a Sentry Relay - the thing that gets the events, filters them and forwards them to Sentry-land - on your localhost. There's a Github repo where you can download a build for your OS.
I've downloaded it and after some fumbling with the configuration got it to work with verbose logging and route my events to my Sentry project. All was well until I've started working on more sophisticated events called transactions.
Even though the events were received by Sentry successfully, the transactions just weren't appearing in the dashboard. I've looked at the Relay logs and in all the verbose mess of trace level logging I've noticed "invalid transaction dropped".
And that's it. No details. Just that.
I've blindly tried to minimize the number of fields I was sending to only the required once. I've tried some changes based on guessing. Didn't help.
I've googled. Of course I did. I've googled the error and through some hops found the place in the Relay's source code it was doing the transaction validation.
Here's the link if you want to follow along:
The code is in Rust and the function is called validate_transaction.
OK. There are checks for having a trace context and for that context to have specific fields. I've had all that in my transactions. But there was also a call to another function - validate_timestamps. And the first check is:
if end < start {
return Err(ProcessingAction::InvalidTransaction(
"end timestamp is smaller than start timestamp",
));
}
Yeah. So I don't know about you, but when I need a timestamp I just call now or something. And the way I called it in my test code was first for the end timestamp and then for the start timestamp, making the time go in the wrong direction.
The two lessons
Lesson one: when you write a software that does validations, please tell the users of that software why validation failed. Don't just say "validation failed". Add the reason.
Lesson two: if the source code is available, jump into it and start digging. It's probably faster that way and unlike documentation it doesn't have missing bits or outdated information. And the more times you do it, the easier it gets to read source code that you didn't write. You actually get some satisfaction from it. Like you've uncovered some knowledge from the source directly. No one spoon fed you. You did it yourself. Have a cookie!