Due to supporting older SDKs and ways of submitting feedback, we have 2 storages. The sentry_userreport table in postgres (attachments dataset), and search_issues in Snuba (shared with other Issue Platform types). Very few services use Postgres and we'd like to stop using it entirely, since Snuba is a superset of all feedback. This will
Reduce inconsistencies, bugs, and developer effort from using 2 storages.
Dedup the data we're storing for this product.
Tasks:
migrate issue details feedback tab to query snuba search_issues instead of postgres. Do the same for all sentry features that use the model (pretty sure it's just issue details).
decide whether to
a. stop supporting the /user-feedback/ api entirely, or
b. call /issues/ under the hood and return extra fields, but ensure backward compatibility
i. don't actually call /issues/, just copy the backend query
make a task to delete already-shimmed user reports. At this point, nothing should be querying these.
Due to supporting older SDKs and ways of submitting feedback, we have 2 storages. The sentry_userreport table in postgres (attachments dataset), and search_issues in Snuba (shared with other Issue Platform types). Very few services use Postgres and we'd like to stop using it entirely, since Snuba is a superset of all feedback. This will
Tasks:
migrate issue details feedback tab to query snuba search_issues instead of postgres. Do the same for all sentry features that use the model (pretty sure it's just issue details).
decide whether to a. stop supporting the /user-feedback/ api entirely, or b. call /issues/ under the hood and return extra fields, but ensure backward compatibility i. don't actually call /issues/, just copy the backend query
make a task to delete already-shimmed user reports. At this point, nothing should be querying these.