Vladislav Shpilevoy

Rebalancer is started on one of cluster nodes, for example, on a master with minimal UUID, as it works in VShard. Rebalancer asks bucket count from different replicasets, calculates a...

It works on a replicaset master and recovers bucket, whose relocation has failed for any reason: network problems, one of nodes denial etc. It is quite trivial. Just checks a...

It is two counters: RO and RW. RO reference counter protects garbage bucket from being deleted while it has some active requests working with it. RW reference counter will protect...

A fiber, scanning partitioned spaces by their bucket_id index and deleting tuples, having bucket_id with GARBAGE status in _bucket space. Garbage collector is started on a replicaset master with the...

It is going to be used by rebalancer and provided as a public API. Just like in VShard. See VShard for examples, API, tests. Follows #3987.

There are three ways: 1. Add an option to index_create: ‘partition_by’ or ‘is_shard_index’, or something like that, having boolean value. True if a first field of this index should be...

Add this space to store meta info about each bucket. For a good example see VShard _bucket. Add options to box.cfg - bucket_count, weight. One of the most non-trivial things...

System fields are similar to rowid in traditional SQL databases, and are not visible via requests like ‘SELECT *’. Nonetheless a system field can be obtained via direct name specification,...

Change ‘cluster’ to ‘replicaset’ in names of some constants, functions etc. For example, rename GROUP_DEFAULT to GROUP_REPLICASET. Further under cluster multiple replicasets are meant. This is a preparation for sharding...

Recently it was allowed to set fields in arrays out of their boundaries. The missing fields were added as nils automatically. But it does not work if the updated array...