For non-200 responses, we generate a Sentry log "Non-200 response hitting references counter API". When the system has a problem, as it did in late March, this floods Sentry with these logs and can quickly deplete our error logging quota (resulting in the dropping of logs until the next monthly billing cycle starts).
I guess we should either consolidate these logs better so that fewer separate events are logged (preferable) or disable sending this particular type of logs to Sentry.
For non-200 responses, we generate a Sentry log "Non-200 response hitting references counter API". When the system has a problem, as it did in late March, this floods Sentry with these logs and can quickly deplete our error logging quota (resulting in the dropping of logs until the next monthly billing cycle starts).
I guess we should either consolidate these logs better so that fewer separate events are logged (preferable) or disable sending this particular type of logs to Sentry.