sunnysabor opened a new issue, #6904:
URL: https://github.com/apache/shenyu/issues/6904
### Is there an existing issue for this?
I have searched the existing issues.
### Current Behavior
When the same Eureka service is configured as a discovery listener node in a
non-default ShenYu namespace, the initial discovery refresh can populate
upstreams with the correct namespace, but subsequent Eureka incremental events
may be filtered out.
The incremental event path currently has a namespace mismatch risk:
- `EurekaInstanceRegisterRepository.buildUpstreamJsonFromInstance(...)`
builds upstream JSON with `url`, `weight`, `protocol`, `props`, and `status`,
but no `namespaceId`.
- `DiscoveryDataChangedEventSyncListener.buildProxySelectorData(...)` parses
that JSON. When `DiscoveryUpstreamData.namespaceId` is blank, it defaults the
upstream to `SYS_DEFAULT_NAMESPACE_ID`.
- The listener then filters upstreams by
`discoverySyncData.getNamespaceId()`.
For a non-default namespace, a blank upstream namespace is converted to the
default namespace and then filtered out.
### Expected Behavior
When a registry event payload does not include a ShenYu namespace, the
upstream should inherit the namespace of the current discovery sync context,
i.e. `discoverySyncData.getNamespaceId()`.
This is consistent with `AbstractDiscoveryProcessor.fetchAll(...)`, which
fills missing upstream namespaces from `proxySelectorDTO.getNamespaceId()`
during the full refresh path.
Payloads that explicitly include `namespaceId` should keep the current
filtering behavior.
### Steps To Reproduce
1. Configure at least two ShenYu namespaces, including one non-default
namespace.
2. In the non-default namespace, configure a `divide` discovery selector
with `discoveryType=eureka` and `listenerNode` equal to an Eureka application
name.
3. Ensure the initial full discovery refresh populates upstream data for
that selector.
4. Add, remove, or update an Eureka instance for that application.
5. Observe that the incremental upstream event payload has no `namespaceId`,
is defaulted to `SYS_DEFAULT_NAMESPACE_ID`, and is filtered out for the
non-default namespace.
### Environment
ShenYu version(s): current master
### Debug logs
Relevant code paths:
-
`org.apache.shenyu.registry.eureka.EurekaInstanceRegisterRepository#buildUpstreamJsonFromInstance`
-
`org.apache.shenyu.admin.discovery.DiscoveryDataChangedEventSyncListener#buildProxySelectorData`
- `org.apache.shenyu.admin.discovery.AbstractDiscoveryProcessor#fetchAll`
### Anything else?
The minimal behavior-preserving fix is to fill missing upstream namespaces
from the current discovery sync context:
```java
if (StringUtils.isBlank(discoveryUpstreamData.getNamespaceId())) {
discoveryUpstreamData.setNamespaceId(discoverySyncData.getNamespaceId());
}
```
This preserves behavior for payloads that explicitly include `namespaceId`,
while making blank namespace handling consistent with the full refresh path.
--
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]
