XnLemon commented on PR #3369: URL: https://github.com/apache/dubbo-go/pull/3369#issuecomment-4642315610
> **Findings** > > * [P1] Standalone `metadata-report` configs can no longer be used by named registries. > In PR head, [GetMetadataReportByRegistry](/tmp/dubbo-go-pr3369.3fkfL9/metadata/report_instance.go:86) returns `nil` for any non-empty unknown `registryId`. But standalone `metadata-report` config is still registered under `default` in [config/metadata_config.go](/tmp/dubbo-go-pr3369.3fkfL9/config/metadata_config.go:91), while registry URLs get ids like `nacos`/`zk`. That means `serviceDiscoveryRegistry.metadataReport` becomes `nil` in [service_discovery_registry.go](/tmp/dubbo-go-pr3369.3fkfL9/registry/servicediscovery/service_discovery_registry.go:81), and remote metadata publish/fetch fails for a common config shape: named registry plus separate metadata-report. Need either preserve `default` as an explicit global fallback or add config/API support to bind standalone metadata-report to registry ids. > * [P2] #3352’s mapping-read problem is still not fixed. > [ServiceNameMapping.Get](/tmp/dubbo-go-pr3369.3fkfL9/metadata/mapping/metadata/service_name_mapping.go:93) still reads from `GetMetadataReport()` only, so in multi-report setups it now deterministically picks `default` or lexicographically first. That makes the old random behavior stable, but it still does not use the subscribing registry’s report, nor does it fan out consistently with `Map`/`Remove`. The issue explicitly calls out “mapping writes all reports, reads one report”; this PR only changes `Remove`. > on is resolved. 第一个修了 目前可以正常fallback 第二个 模型统一 把get改为拿所有report然后遍历查求并集合并结果返回result 中间如果有一个挂了记录错误但不影响继续 只有全挂了才报错 -- 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] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
