Hi All, am using v1DataEngine, its taking 12% cpu from each node. If i have 3 node Kubernetes cluster of 96 cpu each, and if longhorn is on each node, it takes 11.52*3=34.56 only for longhorn. Helm says we can change this 12% with guaranteedInstanceManagerCPU. What is the consequence of reducing guaranteedInstanceManagerCPU? Question: ----------- Why guaranteedInstanceManagerCPU is mentioned in percentage? What is the consequence of reducing guaranteedInstanceManagerCPU to 4% or 5% on an 96 cpu server.?

Hi @late-needle-80860, can you please help with this query posted by my colleague?

The consequence is that Longhorn gets less dedicated shares of the cpu and thereby something else can take those cpu cycles and that can result in Longhorn’s controls and the like to be delayed which in turn can result in a engine and replica crash … meaning a ( hopefully ) temporary loss of volume for x workload that might need it. For more see the docs ( as always ) > https://longhorn.io/docs/1.9.0/references/settings/#guaranteed-instance-manager-cpu