It can be learned from the document that nodetool scrub can identify and repair damaged SSTables. So, under what circumstances does this tool need to be used in a production environment?
Should we use it as a scheduling task and keep scheduling it in the background, or should we use it when troubleshooting problems? Or is it that we only use it when migrating SSTables?
Thanks!
Currently, scrub is meant to be used in response to specific problems found in sstables.
For the future, we plan to improve it to be a general tool which gets rid of bad sstables (not necessarily repairing them – only when possible) and also scheduled job which regenerates sstables periodically.