Fix scale back to 0 with 300 requests scenario

substratusai / lingo

Lightweight ML model proxy and autoscaler for kubernetes

https://www.substratus.ai

Apache License 2.0

96 stars 6 forks source link

Closed samos123 closed 5 months ago

samos123 commented 5 months ago

Fixes #73 and includes pr #70 and others from Alex

Remove compareScales function
Refactor SetDesiredScale that is simpler to understand by storing lastSuccessfulScale time
Improve test coverage for scaler.go

Co-author: @alpe