Closed cjagus closed 4 years ago
The hashing is based on the metric name (not the size/retention of the metric), but in most cases I'd expect the number of wsp files to be really close throughout the Graphite cluster when using the jump hash.
Run bucky inconsistent
and see what it says.
I tried to rebalance a single jump_fnv1a carbon instance by adding two more, initial one was having a disk size of 6gb.
After the rebalance:
instance1 = 2.1gb instance2 = 1.6 gb instande3 = 1.6 gb
Is this an expected behaviour? ( considering jump hash has a much better balancing b/w servers)
I tried below command to do the same.
bucky rebalance -delete -h instance1 :4242 -w 25 2>&1 | tee rebalance.log
[using bucky/buckyd 0.4.0]