You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
We currently never use neonvm-daemon /cpu GET endpoint, since it has no support for fractional CPU limits. Instead we rely on the value stored in the runner memory via atomic variable, which is updated on each scaling request.
As a first step for fractional limits, we might want to move that variable into the neonvm-daemon.
Reasons:
as said, first step towards proper fractional limits
clearer ownership - all cpu related state will be kept in a single component (neonvm-daemon) rather than in 2 components as now (neonvm-daemon, neonvm-runner)
We currently never use neonvm-daemon /cpu GET endpoint, since it has no support for fractional CPU limits. Instead we rely on the value stored in the runner memory via atomic variable, which is updated on each scaling request.
As a first step for fractional limits, we might want to move that variable into the neonvm-daemon.
Reasons:
Context:
Discussion in #1190 (comment)_
The text was updated successfully, but these errors were encountered: