This message was deleted.

Can you provide more information about the engine pod resource reduction you mentioned? Is the engine means longhorn-engine? https://github.com/longhorn/longhorn-engine

Hi Ray！ The engine pod is instance-manager-e***. It includes tgtd process and engine controller.

Got it, I might think you can refer to here to adjust the resource that allocate to engine pod.

Yeah~ I am trying to adjust the resource.But it is useless.At the same time, I find When I start a vm, disk I/O is very smooth. In contrast, disk I/O performance degrade.It's only 1/3 of what was expected.

Hi @bright-wall-24452, thank you for response. I would like to ask further information about the context. • Are the I/O performance benchmark results you mention compared to your local disk? • Does the CPU/Memory reach the upper bound?

Yeah~ Thank you for your response. The I/O performance benchmark results are the local disk. That is ISCSI. The CPU/Memory reach Approximately 300%. My host has 80 cores. I found when my host has 256 cores，The performance of I/O has been improved.

@bright-wall-24452 Sure, it is very welcome to have contribute from community. i think you can create a issue ticket in https://github.com/longhorn/longhorn/issues, then having a pull request to improve that.