ClusterID has been saved in too many places

[okJiang]

Enhancement Task

ClusterID is the ID which is initialized before startServer. It is initialized once during the serving.

ClusterID init

pd/server/server.go

Line 436 in 8571ca4

s.clusterID.Store(clusterID)

After this, all operations on ClusterID are read operations. To read it, PD passed the ClusterId layer by layer and saved it.

pd/server/server.go

Line 493 in 8571ca4

s.cluster = cluster.NewRaftCluster(ctx, clusterID, s.GetBasicCluster(), s.GetStorage(), syncer.NewRegionSyncer(s), s.client, s.httpClient)

pd/server/server.go

Line 503 in 8571ca4

s.keyspaceGroupManager = keyspace.NewKeyspaceGroupManager(s.ctx, s.storage, s.client, clusterID)

pd/server/server.go

Line 507 in 8571ca4

s.hbStreams = hbstream.NewHeartbeatStreams(ctx, clusterID, "", s.cluster)

pd/pkg/mcs/metastorage/server/manager.go

Line 51 in 8571ca4

m.clusterID = srv.(T).ClusterID()

pd/pkg/mcs/scheduling/server/meta/watcher.go

Line 61 in 8571ca4

clusterID: clusterID,

pd/pkg/mcs/scheduling/server/cluster.go

Line 100 in 8571ca4

clusterID: clusterID,

Strongly related to the ClusterID is the rootPath field, which is also passed and stored hierarchically.

pd/server/server.go

Line 442 in 8571ca4

s.rootPath = endpoint.PDRootPath(clusterID)

For ClusterID, a better way is to use a global variable so that we can clearly know where it was modified and where it was used.