investigate edge cases, where queried neighbour node (when joining or leaving) is own VS
Let X be a neighbour node, O an active vserver to be relocated and `
investigate edge cases, where queried neighbour node (when joining or leaving) is own VS
Special cases of the edge cases above: The own vs queried is the one that's departing/ relocating.
investigate edge cases, where queried neighbour node (when joining or leaving) is own VS
v-server change needs to trigger evaluation and persisiting of load stats
extend DHT protocol with load information sampling request
k-choices: manage v-servers
periodic k-choices rebalancing
update node ID validity check to allow different vserver IDs
update node ID validity check to allow different vserver IDs
Turns out that no interval checking needed to be done, because each RemoteNodeState sent with every request inludes the ID of the sending vserver. So no trying out of all the vservers is needed.
Still did a bit of cleanup and refactoring ¯_(ツ)_/¯
adjust
isJoined to kchoices semantics
interface between DHT and Service to request load information
try all possible vservers when bootstrap joining