clicksync

Tool for synchronizing clickhouse clusters

works only with partitioned MergeTree tables
can sync clusters with different node number
uses info from system.clusters to determine nodes
creates tables on destination cluster
recreates tables if schema differ
checks record number in each partition each run, and refill if numbers differ (sync)
can work in two modes: (a) suck data in and put it to destination cluster (very slow) and (b) direct, using insert ... select from remote()
can copy only recently created partitions

usage

adopt clicksync.yaml as you need, then run

clicsync.py clicksync.yaml