This message was deleted.

I thought

druid.coordinator.period

is the config that controls when the coordinator decides to balance out segments. I haven't set it, so assuming the default to be 60s, shouldn't the load queues start getting loaded?

I thought the segments would eventually balance themselves out across all nodes, so the new node would eventually get some ... but I understand this can take a while longer once the priority task of loading all replicas of segments somewhere is complete.

Not sure when this happened, but the data is now evenly distributed among all the historical nodes. Noticed this only this morning!

Hi Ankur, that's what I've heard from others who have experienced the same issue ... balancing is lower priority than initial loading, so sometimes it takes a while for Coordinator to get around to handling it. Just curious, how many segments in your cluster? It's possible if the number of segments is large that would contribute to the amount of time it takes to run through this process.

there are close to 75k segments right now

Okay that's not an unusually large number. Unless you are running with a very small cluster, and/or have very small coordinator/master nodes (i.e. < 4 cores)

well not really, all our servers run on servers with resources equivalent to i3x4 Large

Okay then there is probably no issue, just a matter of time until Coordinator decides to start segment balancing. Your original post was after waiting only 10 minutes ... do you recall if you checked further after that? 30 minutes, or an hour later?

good point, no I didn't. I only saw it in the morning today. Let me see if I get some task info from postgres (metadata RDBMS)