I ended up with over-replicated blocks which are not getting deleted..
I did like following..
Started hadoop cluster with three DN's
Written 1k files with RF (Replication Factor) =3
Change RF =2 and Exclude one DN from cluster using decommission
After decommission successful again include same DN(which is excluded) to cluster.(by removing entry in exclude file and execute refreshnode)
In UI I am able to see RF=2 but in fsck report shown as RF=3 and all are over-replicated blocks.
i) I am not getting why NN is not issuing delete command for over-replicate blocks.?
ii) why fsck and ui are showing different RF for same file?
Please correct me if I am wrong...
If it is issue I'll file?
Thanks And Regards