Idea #16442
Updated by Tom Clegg over 1 year ago
Objectives, proposed architecture, and implementation notes: [[arvados:Efficient Main things to address are * live access logs do not scale well (lots of RailsAPI / database / network load when many containers are running, even if no clients are even looking at the logs) * when crunch-run or the compute instance dies, last few log entries are a good place to container logs]] look for clues, and they are lost