====== Placement Groups ======
PG


===== Calculations =====

  * [[https://old.ceph.com/pgcalc/]]
  * Max PG per OSD is 300
  * Result must be rounded up to the nearest power of 2.

===== reweight =====

Adjust reweight according to current OSD utilisation - to prevent filling one OSD
<code bash>ceph osd reweight-by-utilization</code>

In CRUSH hierarchies with a smaller number of OSDs, it’s possible for some OSDs 
to get more PGs than other OSDs, resulting in a higher load. You can reweight 
OSDs by PG distribution to address this situation by executing the following:

<code>
osd reweight-by-pg [<oload:int>] [<max_change:float>] [<max_osds:int>] [<pools>...]            reweight OSDs by PG distribution [overload-percentage-for-consideration, default 120]
</code>

<code bash>ceph osd reweight-by-pg 101 0.05 999</code>

NOTE: 
  * 101 is a ''oload'' parameter, which means ''overload percentage for consideration 101%''. For example, an argument of 200 would reweight OSDs which are twice as utilized as the average OSD.
  * 0.05 is a ''mach_change'' - maximum modification of weight
  * max_osds - maximum number of OSD to be modifies (default is 4)





===== Balancer =====

When the cluster is healthy, the balancer will incrementally move a small fraction of unbalanced PGs in order to improve distribution. This fraction will not exceed a certain threshold that defaults to 5%.
Set it to 1%:
<code bash>ceph config set mgr target_max_misplaced_ratio .01</code>