The single most time-intensive database query right now belongs to the transcribot queue query, taking roughly half a second to process and called every two seconds. This PR does two things:
drastically limits the data that is actually pulled from the database
bypasses the serializer, which chops off a fair amount of extra processing we don't need
We will not know how much of a difference this makes until this is deployed and we can observe with honeycomb, but my gut reaction is that we should see a significant change.
Relevant issue: N/A
Matching tor_ocr PR: https://github.com/GrafeasGroup/tor_ocr/pull/40
Description:
The single most time-intensive database query right now belongs to the transcribot queue query, taking roughly half a second to process and called every two seconds. This PR does two things:
We will not know how much of a difference this makes until this is deployed and we can observe with honeycomb, but my gut reaction is that we should see a significant change.
Checklist: