The invention discloses a method for operating a multi-CPU (Central Processing Unit) architecture service in Kubernetes. The method specifically comprises the following steps: S1, a Kubernetes component is supported and deployed on an AMD64 architecture by default; s2, according to a CPU architecture which is deployed in a Kubernetes cluster and is to be adapted to a service, establishing a service cluster, a container mirror image of ARM64 and a container mirror image of AMD64 are manufactured respectively. S3, customizing the name of the server mirror image by the container cloud platform according to a specified format; s4, according to the mirror image name and the adaptive CPU architecture type, obtaining a mirror image; the method comprises the steps of S1, establishing a containercloud platform, S2, establishing a container mirror image database integrated with the container cloud platform, S3, establishing a container mirror image database integrated with the container cloudplatform, S4, establishing a Docker manifest file, and pushing the Docker manifest file to the container mirror image database integrated with the container cloud platform, and S5, carrying out container mirror image management on a multi-CPU architecture through a container cloud platform mirror image management function. According to the method for operating the multi-CPU architecture service inthe Kubernetes, the multi-CPU architecture service is operated in the Kubernetes; according to the technical scheme, the requirement for daily use can be met through improvement, the effect of supporting management of multiple CPU architecture hosts in the Kubernetes cluster is achieved, and therefore the Linux containers of the corresponding CPU architectures can be operated on the correspondinghosts.