Tracking issue for Next-Gen TiDB

[rleungx]

Development Task

This issue is used to track the development or bug fix progress of PD when adapting to the new architecture of TiDB.

Keyspace

• next-gen: add reserved keyspaces #9414
• Validate the keyspace name according to tidb_service_scope #9348
• Return the cause error when a keyspace is not exist in the client SDK #9704

Resource control

The status of the resource control is tracked separately by

• Tracking issue for the Keyspace Service Limit #9296

GC

The status of the new GC API is tracked separately by

• Refactor GC related PD APIs #8978

Misc

• Introduce a build tag for next-gen #9350
• [next gen on clould] evict-slow-store-scheduler is enabled even though evict-slow-store-scheduler is not supported #9629

Bug fix

• [next gen on clould] AP query report error: resource group default not found, maybe has been deleted #9448
• [next gen on clould] add new keyspace failed with error “[PD:keyspace:ErrKeyspaceExists]keyspace already exists\""]” #9355
• The keyspace id will be allocated over the uint32 max #9417
• [next gen on clould] some pd metric data are abnormal #9652
• [next gen on clould] The oversized count is not accurate #9640
• TSO/scheduling server has wrong TLS setting #9367
• [next gen on clould] native br restore failed due to scatter failed #9482
• [next gen on clould] pd oom during tikv scaling up #9555
• [next gen on clould] gc interface can blow up the PD during simulating one of pd follower network delay 100ms #9724
• [next gen on clould] workload report " FATAL: mysql_drv_query() returned error 8252 (Exceeded resource group quota limitation) for query 'INSERT" when simulate pd leader network partition #9727
• [next gen on clould] pd metrics abnormal after simulated pd leader network partition #9729

Testing

• Stabilize nextgen testing #9806