I'm running a Cassandra 2.2.8 cluster of 3 nodes in a single data center. The replication factor is 3.
I had to recover one of the nodes from an old backup. After recovering the backup I ran a full repair to sync the latest data, but it seems that the recovered node has less data than the others:
How can I force data synchronization?
One idea might be checking the system.log when doing full repair. is there any error over there about repair? if so, debug the errors and fix them.
Otherwise, one pre-check is to run nodetool verify on the recovered node to see whether there're any corrupt sstable even after repair?
If nothing is wrong, there're a lot of possibilities and it's hard to debug without more knowledge on your cluster. Cassandra also doesn't have a direct idea about whether all data are consistent.
maybe if you know the number of object you already put, you can run cqlsh command to count all data tables to see whether they're a match to see whether you really loose any data?
