Different number of objects in master and prole member

[Ivan44785372]

Could you please explain how it is possible. I don’t have pending migrations and at the same time the number of objects in master and prole is different. I thought it should be equal after all migrations is done.

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~Namespace Object Information (2019-03-14 06:19:57 UTC)~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Namespace      Node                Total     Repl                           Objects                   Tombstones             Pending   Rack
    .                             Records   Factor        (Master,Prole,Non-Replica)   (Master,Prole,Non-Replica)            Migrates     ID
    .                                 .           .        .                                 .                            .             (tx,rx)      .
namespace      node1:3000           1.020 B   2        (629.228 M, 391.145 M, 0.000)     (0.000,  0.000,  0.000)      (0.000,  0.000)     0
namespace      node2:3000           1.020 B   2        (391.144 M, 629.229 M, 0.000)     (0.000,  0.000,  0.000)      (0.000,  0.000)     0
namepsace                           2.041 B            (1.020 B, 1.020 B, 0.000)         (0.000,  0.000,  0.000)      (0.000,  0.000)

[meher]

Well, the number of master and prole on a given node is not supposed to match. The sum of all masters and all proles across a cluster should match and they seem to in your case. Having said that, it does seem the distribution of partitions is pretty skewed. This is not necessarily unexpected when not using the prefer-uniform-balance configuration option (available on Enterprise Edition only). On a 2 node cluster running replication factor 2, each node will own all the data anyways.

[kporter]

Prefer uniform balance is generally not needed in really small clusters. This could be caused by eviction which can favor one node in a two node cluster (this is one of the issues that the new NSUP addresses in 4.5.1.5).