Even when one table is corrupt your repair will fail. To handle this case
without data loss you could replace the complete node ( safest and sureshot
option)
On 19 May 2020 20:20, "Leena Ghatpande" wrote:
> One of the tables SStable got corrupted on all nodes. But repairs were
> failing for
One of the tables SStable got corrupted on all nodes. But repairs were failing
for all the tables in the keyspaces.
So I took the cluster down and did an offline sstablescrub and that fixed the
corrupt table, but had loss of data. I did not get a chance to try the
consistency_all option. will