negasus/rplx

Golang library for multi-master replication integer variables with TTL

rplx's Issues

Add prometheus metrics

Add prometheus metrics

• count of incoming/outcomping sync requests
• size of sync requiests
• count of remote nodes
  etc.

Possible write to closed channel

While closing channel in node.go#105 close(n.syncQueue), possible write to channel in node_sync.go#16 case n.syncQueue <- struct{}{}:

Implement compacting values from not exists nodes

Save memory proposal.

If many nodes leaved cluster (and possible joined to cluster with new ID), its values with old IDs will be take memory forever.

Consider compacting proccess for orphans values to one virtual node value.

For example. Node2 and greater are orphans

Var1:
node1: 100
node2: 200
node3: 300
node4: 400

After compacting

Var1:
node1: 200
SpecialVirtualNode: 200+300+400