Do partner1 and partner2 in mover BackrubDD mean chain1 and chain2 respectively?

I was doing the design of a protein interface. The starting structure was two chains in contact with each other.

Previously I have tested turning on and off partner1 and partner 2 of BackrubDD, so what I understood from the design behavior was:

partner1="0" -> no flexible design of chain 1 backbone
partner1="1" -> with flexible design of chain 1 backbone
partner2="0" -> no flexible design of chain 2 backbone
partner2="1" -> with flexible design of chain 2 backbone

However, in a recent design, I discover even if I turned off backrub of partner2 (i.e. partner2="0"), there was a backbone movement of chain 2 after design.

After comparing the rmsd of C alpha of both chains after and before design, a minority of designs reached an rmsd as high as ~4-5A compared with the starting structure (virtually the whole decoy completely deviated from the original structure).

I have two questions:
1) Do partner1 and partner2 in mover BackrubDD mean chain1 and chain2 respectively?
2) Have you ever seen backrub could move the backbone so the design deviates a lot from the starting structure (e.g. 4-5A Ca rmsd)?

Thank you very much for your help!