From the view point of distributing disk accesses, I changed mount sources on k1nds2 and k1nds3. k1nds2 and k1nds3 read minute_raw(_archive) from stream#0 on k1nfs1 and from stream#1 on k1nfs2, respectively. I also changed a mount option from 'rw' to 'ro' on these two NDS to avoid accidental deletion.

There was no change on k1nds0 and k1nds1. Though a mount option on them should be also changed same as k1nds2 and k1nds3, it interrupts various activities using NDS. So I will do it on the next maintenance day.

Final situation of data access is as follows

node	gwf	minute_raw	minute_raw_archive
k1nds0	hyeades-0	k1tw0	tw0 data on k1nfs1
k1nds1	hyeades-2	k1tw1	tw1 data on k1nfs2
k1nds2	Kashiwa	k1tw0	tw0 data on k1nfs1
k1nds3	hyeades-2	k1tw1	tw1 data on k1nfs2