Background reclaim - Time used by Instance and Time spent in task quota violation by instance are high in few nodes. Is that means those nodes has bad memory or what could be the reason?
This means that you have high memory pressure. ScyllaDB tries to maintain some reserve memory. When this runs low (below some threshold), the background reclaim kicks in to free some reclaimable memory. If you have high CPU usage in this scheduling group, it means all easily reclaimable memory was already reclaimed and the reclaimer resorted to expensive LSA memory compaction.
Check your memory metrics, you probably have high non-LSA memory consumption. A typical cause of this is high bloom-filter memory usage.