Master node network bandwidth requirements

Hello, I have a cluster with dedicated master nodes. I recently noticed, that the current master has relatively high network bandwidth requirements (sustained non-stop ~15Mb/s in, and ~15Mb/s out). It seems that the traffic between the master node and the individual data nodes is proportional to the number of indices/segments on the respective data node. Is there any info/guide regarding the network requirements of master-only nodes and how to optimize it?

Also, as a bonus, I noticed a long-term pattern, where the traffic goes rapidly down (around 50%) every 3-4 days and then slowly climbs back up to the ~15Mb/s over the course of 24 hours, stays there for ~3 days and falls back to 50% as the next "iteration" starts. Any idea what could be the reason? The number of indices is only growing, we do not delete large amounts of data with that period.

That traffic pattern doesn't match anything master-related of which I'm aware. Possibly GET _tasks?detailed will give some clues about what it's doing. If not, there's some more detailed network stats coming in 8.8.0 which will answer this for sure:

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.