Show how to mutate the FluentBit Daemonset pods resources.requests to fit the expected load that depends on node size.
Why is this needed?
Daemonset log and metric forwarders (e.g., Fluentbit) require CPU and memory resources proportional to node size.
In largenodes (e.g., 16xl vs. xl), setting resources.requests could hinder log and metric forwarders’ performance. Alternatively, over allocation of FluentBit resources.requests pods to fit the most extreme case (the largest instance size) is wasteful.
What would you like to be added?
Show how to mutate the FluentBit Daemonset pods
resources.requests
to fit the expected load that depends on node size.Why is this needed?
Daemonset log and metric forwarders (e.g., Fluentbit) require CPU and memory resources proportional to node size.
In largenodes (e.g., 16xl vs. xl), setting
resources.requests
could hinder log and metric forwarders’ performance. Alternatively, over allocation of FluentBitresources.requests
pods to fit the most extreme case (the largest instance size) is wasteful.