Spike sometimes reduce set reads


#1

AS randomly have a low TPS as saw in screenshot http://joxi.ru/KAg7K3wcwVPOAl In logs I have nothing about it. Can anybody help to undertand how I can find a cause of this. On system side (memory / CPU / IO) all looks fine.


#2

The length of these dips seem very consistent, ~2 min.

A few questions:

  1. Could you include logs from an affected node?
  2. What version of the server are you using?
  3. Which client are you using and what is the client’s version?
  4. Are you using an asynchronous client?
  5. Were there migrations happening?
  6. Was these events preceded by the cluster changing the cluster size?
  7. Are you running on your own hardware, if so are you using virtualization?
  8. Are you running on hosted hardware and if so where is it hosted?