Ceph

<vikhyat> kefu: --no-increasing are you aware of documenting this option
<vikhyat> I could not find much
<vikhyat> in reweight-by-utilization
<kefu> vikhyat, it not documented anywhere other than in c++ =)
<vikhyat> :P
<vikhyat> can you please help me to understand how it helps
<kefu> sure.
<vikhyat> then I will send a doc patch
<kefu> i should have update the ceph.8 last time
<kefu> some guy asked me the other day.
<kefu> oh, many thanks!
<vikhyat> https://github.com/ceph/ceph/pull/8026/commits/06a2a7596b876a5c2addb184a48222bcb6e23b9a
<vikhyat> ahh this one
<vikhyat> np :) as soon as I will learn will send a doc PR
<vikhyat> :)
<kefu> --no-increasing is off by default. so increasing the osd weight is allowed using the reweight-by-* or test-reweight-by* command.
<vikhyat> kefu: okay
<vikhyat> means reweight-by-* or test-reweight-by* commands may increase the weight of some of the osds
<vikhyat> correct
<vikhyat> ?
<kefu> and i assume you understand how r-by-* works?
<kefu> right.
<vikhyat> yes :)
<vikhyat> and if we specify
<vikhyat> --no-increasing
<vikhyat> it will never increase
<kefu> yes, even the osd is under utilized.
<vikhyat> and will do other job of reweight-by-* or test-reweight-by* commands
<vikhyat> right
<vikhyat> So if a user just want to reduce the weight of higher utilized osds
<kefu> vikhyat, maybe the idea of this option is to prevent the cluster from more data migration.
<vikhyat> and do not want to increase the weight of less utilized OSDs they can use this option
<vikhyat> right
<vikhyat> kefu: 100% true
<kefu> so we can increase the # of changed osd, enable increase , step by step.
<vikhyat> kefu: sending a small doc pr for ceph.rst
<kefu> heading towards a more balanced cluster

https://github.com/ceph/ceph/pull/9144