These are chat archives for cern-eos/eos/operations

20th
May 2016
lmascett
@lmascett
May 20 2016 07:04
@davidjericho in the scheduling group default.11 you have ongling 954 balancing jobs (inside that space)
@davidjericho if you run as well eos node ls --io you will see how many are scheduled per node and if you run eos io stat -x you will see the statistics od data transfers for the balancing application
depending on how much aggressive you want the balancing to be you can increase/reduce the amount of balancer.node.ntx (which is a node wide setting)
obviously this depends as well on how much you machine can handle and how small/big are your files
in general our instances are configure to do a gently balancing (between 2 and 10 transfers per node) in order not to overload the possible destination
lmascett
@lmascett
May 20 2016 07:10
@davidjericho if you run this we might have an idea about your curranet file/space distribution in your scheduling group eos fs ls --io | grep default.11