1. There are probably a few hairs left on the create table stuff. I've worked out those that I needed to make my own test cases work, but it's possible I've missed something. Postgres allows lots of qualifiers on table rows, so they're interesting to handle. If you can find an error statement in your logs, let me know and I'll fix it.

2. The scheduled sync feature of dbreplicator is left over from daffodil replicator. I've been meaning to go back and fix it. Essentially, real-time replication just continuously tries to sync. Errors were not handled very well in daffodil, and I'd had the same problem. I wasn't looking for real-time replication, so I implemented the daemon process for linux/unix systems. I could go back to the scheduler and fix it so it ignores a certain number of a certain type of errors, such as 'blocked by another user', which essentially means that another subscriber is currently syncing. What I'd meant to do was implement threading on the publisher process so that subscribers could simultaneously sync. Right now it's synchronous, meaning that only one subscriber at a time can be syncing. If you try to sync while the publisher is busy with another subscriber, it sends back the error you saw.

~b