Ask HN: Anybody using postgres-xl?

I have recently started using postgres-xl and I'm quite impressed with it although there is one problem somebody flagged on stack-overflow that makes me worried - the sharded database cannot be replicated to avoid downtime when one of, say, coordinators or data nodes goes down.

So I started testing some ideas on using keepalived to elevate live fail-over without much of a downtime and without my intervention at the time of fail-over. The intention is to document it and share back to the community. I am through two weeks of night-testing and now I have hit the wall I cannot find the solution myself.

In postgresql the one can bind the process to IP interface. Coordinators and data nodes are just postgresql processes, though they seem to ignore listen_address option. The process itself can be started with extra parameters (-h ip_address) which will result in what I need however postmaster.opts file (which seems to be used by pgxc_ctl to start coordinator) gets overwritten by pgxc_ctl each time coordinator is started (before it's started).

Does somebody from YC community have some experience with this?