We have Cassandra-0.8.2 cluster of 24 nodes and replication factor 2 . One of the node is quite slow and most of sstables on this node is corrupt.(We are not able to run compaction and not even scrub)
So is it possible to clean the data,cache and commitlog directories for this node and restart with bootstrap=true? Will it help to get all the data stream back to this node?
If it is possible , is there anything that could create issue?What care should be taken to avoid any danger?
As long as you have your replication factor set to 2. you should not have a problem to clean up and restart the machine node. But it would take some time for the data to flow through, Mine took around 4 hours. A Best way to analyse this visually is to install Opscenter from the DataStax. Its a great tool. There is no danger. Let us know if you were able to succeed!
Also it is advisible to upgrade to Cassandra 1.0. it is much faster! & you will instantly notice the difference.