I have started repairing a 10 node cluster with one of the table having > 1TB of data. I notice that the validation compaction actually shows >3 TB in the "nodetool compactionstats" bytes total. However, I have less than 1TB data on the machine. If I take into consideration of 3 replicas then 3TB makes sense. Per my understanding, validation does only care about data local to the machine running validation compaction. Am I missing some thing here? Any help is much appreciated.
Thanks,Sri