Reputation: 1
While using Cassandra 2.1.17, I encounter a problem in this scenario:
Sometimes I can get lot of extra data in local cassandra cluster which can not be found in the online cluster.
I pick one of these records, using the sstable2json to explain the sstables, I can find a tombstone record with the exactly key in the file where I download from s3 directly, whose timestamp is greater than the record's. That means this record has been deleted and for some reason, it revived during restore to the local cluster.
What's more, I search the local cluster's data, only the data record is still here while I can't find the tombstone record.
Have anyone met this before? Is't a sstableloader's bug? If so, how can I prevent it?
Upvotes: 0
Views: 66
Reputation: 361
Just wondering about how well time might be synchronized between prod cluster and local node and if this could influence on TS revival...
Upvotes: 0