How can we recover sstable if it is corrupted or deleted at the OS level?
How can we recover a SSTable if it is corrupted or deleted at the OS level?
If an sstable is deleted it cannot be recovered unless you have a backup. Alternatively, if this cluster is replicated, meaning you have a replication factor greater than 1, then you could just run a repair on that table and the sstable will get regenerated. If you have an RF of 1 only, then there's no alternative but to restore from backup/snapshot.
Thank you Steve.
What can we do if sstable is corrupted?
We have RF 3
You have a few options with an rf=3:
1) delete the sstable and repair the table
2) nodetool scrub
3) sstablescrub
Nodetool scrub is an online tool, so you can give that a try and see if the stdout reports anything. Alternatively, you can take the node offline and then run sstablescrub which is much more thorough. Lastly, and easy enough is just remove the sstable that's causing a problem and run a repair on the table which will regenerate the missing data.
One note with nodetool repair, be sure you haven't repaired the table with incremental repair in the past. To do so, I would run nodetool mark_unrepaired first, then run nodetool repair.
7 People are following this question.
