ref(eap): Don't write sentry.browser attributes#5914
ref(eap): Don't write sentry.browser attributes#5914loewenheim wants to merge 1 commit intomasterfrom
sentry.browser attributes#5914Conversation
Relay writes the attributes `sentry.browser.name` and `sentry.browser.version` in user agent detection. However, the official names of these attributes in conventions are actually `browser.name` and `browser.version`. The `sentry.` variants have been declared as aliases of the correct versions in `sentry-conventions` for some time, and Snuba is able to coalesce them on this basis, so no part of the product should require the `sentry.` variants to work. Therefore, it is safe to write the correct versions of these attributes.
| # Calculated byte size: name + value + attribute keys/values. | ||
| # This is a billing relevant number, do not just adjust this because it changed. | ||
| "quantity": 249, | ||
| "quantity": 235, |
There was a problem hiding this comment.
This does give me a little bit of pause; we're making a purely internal change here (the names of some attributes which we infer, not anything that the "user" sends) and it has an effect on a billing-relevant number. Is it correct that we take the names of attributes the user didn't send into account for billing?
There was a problem hiding this comment.
Is it correct that we take the names of attributes the user didn't send into account for billing?
I believe so, yes (but with the new ingest_settings the user should be able to opt in or out of these attributes).
My 2c is that this is fine as long as the byte size does not increase. @k-fish what's your opinion?
There was a problem hiding this comment.
These changes must not have an impact on that number. The size logic should be built in a way, that the first (trusted) Relay, counts the original payload size of a trace metric/log and this is then the value used as the billed value.
If an external/untrusted Relay makes these changes, all we can do is consider it an SDK (untrusted client) and we use the size our infrastructure received.
There was a problem hiding this comment.
These changes must not have an impact on that number. The size logic should be built in a way, that the first (trusted) Relay, counts the original payload size of a trace metric/log and this is then the value used as the billed value.
Right, that's what I thought. I'll see if the sizing logic needs to be fixed for trace metrics.
There was a problem hiding this comment.
Fixed in #5917. Fortunately it turns out that only the test was wrong, the logic is working as intended.
3 participants
Relay writes the attributes
sentry.browser.nameandsentry.browser.versionin user agent detection. However, the official names of these attributes in conventions are actuallybrowser.nameandbrowser.version.The
sentry.variants have been declared as aliases of the correct versions insentry-conventionsfor some time, and Snuba is able to coalesce them on this basis, so no part of the product should require thesentry.variants to work. Therefore, it is safe to write the correct versions of these attributes.ref: INGEST-882.