ChuanFF opened a new pull request, #12634: URL: https://github.com/apache/apisix/pull/12634
### Description Currently, the Kubernetes service discovery implementation in APISIX incorrectly handles EndpointSlices, treating each individual slice as the complete set of endpoints for a service. This contradicts the official Kubernetes documentation. According to the [Kubernetes EndpointSlice documentation](https://kubernetes.io/docs/reference/kubernetes-api/service-resources/endpoint-slice-v1/): > "For a given service there may be multiple EndpointSlice objects which must be joined to produce the full set of endpoints; you can find all of the slices for a given service by listing EndpointSlices in the service's namespace whose `kubernetes.io/service-name` label contains the service's name." The correct behavior should be to combine all EndpointSlices sharing the same `kubernetes.io/service-name` label to form the complete endpoint set for a service. ### Solution Approach #### 1. Add EndpointSlices Cache - Implemented `endpoint_slices_cache` to store all EndpointSlices with the structure: `endpoint_slices_cache["k8s_id/namespace/k8s_service_name:port_name"] = {slice1, slice2, ...}` - Any changes to EndpointSlices trigger updates to this cache - The cache is then used to assemble the complete node list for each service, which is subsequently updated in shared memory #### 2. Code Refactoring - Renamed variables to better align with Kubernetes EndpointSlices data structure terminology - Improved code readability and maintainability #### 3. Some Bug Fixes - **Fixed incorrect handling of `addresses` field**: The `addresses` field contains a list of IP strings, not objects with `.ip` attributes - **Optimized node sorting logic**: Simplified the nested loop structure in `endpoint_buffer` processing: ```lua -- Before (incorrect): for _, ports in pairs(endpoint_buffer) do for _, nodes in pairs(ports) do core.table.sort(nodes, sort_nodes_cmp) end end -- After (correct): for _, nodes in pairs(endpoint_buffer) do core.table.sort(nodes, sort_nodes_cmp) end ``` The `endpoint_buffer` uses ports as keys and node lists as values, requiring only a single iteration level. ### Checklist - [ ] I have explained the need for this PR and the problem it solves - [ ] I have explained the changes or the new features added to this PR - [ ] I have added tests corresponding to this change - [ ] I have updated the documentation to reflect this change - [ ] I have verified that this change is backward compatible (If not, please discuss on the [APISIX mailing list](https://github.com/apache/apisix/tree/master#community) first) <!-- Note 1. Mark the PR as draft until it's ready to be reviewed. 2. Always add/update tests for any changes unless you have a good reason. 3. Always update the documentation to reflect the changes made in the PR. 4. Make a new commit to resolve conversations instead of `push -f`. 5. To resolve merge conflicts, merge master instead of rebasing. 6. Use "request review" to notify the reviewer after making changes. 7. Only a reviewer can mark a conversation as resolved. --> -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
