wuyipei opened a new issue, #37738:
URL: https://github.com/apache/shardingsphere/issues/37738

   Issue Title (标题)
   [Feature Request] Support partial initialization: Allow 
ShardingSphereDataSource to start even if single physical data source is 
unavailable
   
   Issue Description (问题描述)
   
   Is your feature request related to a problem? Please describe.
   Yes. In a sharding scenario with a large number of physical data sources 
(e.g., 20+ shards), the stability of the entire system is heavily impacted by 
the connectivity of a single node.
   
   My Scenario:1.I am using ShardingSphere-JDBC in Standalone mode.2.I have 
configured 20 physical data sources (e.g., ds_0 to ds_19).3.Due to network 
jitter or maintenance, only one data source (e.g., ds_15) is temporarily 
unreachable or misconfigured.4.When I initialize ShardingSphereDataSource, the 
entire process fails and throws an Exception.
   
   Current Behavior:
   The initialization is "All-or-Nothing". One failure causes the whole 
application to crash/fail to start.
   
   Expected Behavior:
   I hope ShardingSphereDataSource can support "Graceful Degradation" or 
"Partial Initialization".1.The DataSource should start successfully even if 
ds_15 is down.2.Queries routing to healthy data sources (ds_0...ds_14, 
ds_16...ds_19) should work normally.3.Queries routing explicitly to the failed 
source (ds_15) can throw an exception at runtime.
   
   Why this is important:
   In a production environment, ensuring 100% availability of all 20+ DB 
instances at the exact moment of application startup is difficult. A single 
point of failure shouldn't prevent the service from handling requests for other 
shards.
   
   Config References (Optional):Mode: Standalone / JDBCPool: HikariCP
   
   


-- 
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]

Reply via email to