Details
-
Improvement
-
Status: Open
-
Low
-
Resolution: Unresolved
-
None
Description
In our production clusters, we see several corrupted files on C* severs some times. And we use `Nodetool scrub` to rebuild the tables.
However, out of the 1000+ sstables, usually, just a few are corrupted (less than 10). It will be useful to enhance the `Nodetool scrub` to only scrub certain sstable files. So it will be much efficient than rebuilding the whole table.
One engineer in my team is working on this.