[CapMan visibility] Separates rejecting policy and throttling policy in Sentry tags and spans

[xurui-c]

It's possible for a query to be in the throttling zone for an allocation policy in Snuba, and also in another allocation policy's rejection zone. In which case we see the following:

How could the throttle_threshold be higher than the rejection_threshold? This is due to the query being throttled by an allocation policy with a throttle threshold of 292000000000, but it's also rejected by ConcurrentRateLimitAllocationPolicy The mismatch is caused by not separating out the rejection policy and the throttle policy when creating Sentry tags and spans.

[sentry-io]

🔍 Existing Issues For Review

Your pull request is modifying functions with the following pre-existing issues:

📄 File: src/sentry/utils/snuba.py

Function	Unhandled Issue
_bulk_snuba_query	RateLimitExceeded: Query on could not be run due to allocation policies, info: {'details': {'ConcurrentRateLimitAllo... ... Event Count: 798
_bulk_snuba_query	RateLimitExceeded: Query on could not be run due to allocation policies, info: {'details': {'ConcurrentRateLimitAllo... ... Event Count: 225
_bulk_snuba_query	RateLimitExceeded: Query on could not be run due to allocation policies, info: {'details': {'ReferrerGuardRailPolicy... ... Event Count: 118
_bulk_snuba_query	RateLimitExceeded: Query on could not be run due to allocation policies, info: {'details': {'ConcurrentRateLimitAllo... ... Event Count: 36

---

_{Did you find this useful? React with a 👍 or 👎}