Elvis Pranskevichus

> @elprans, what is the strategy for this command? Is it "create tmp database; reconnect; drop the target one; create target one reconnect; drop tmp database"? Yeah. Ideally, we should...

Does this happen consistently or intermittently? What does `host localhost` show?

This seems like a problem with your system. `host localhost` cannot resolve it either. Check if it's in `/etc/hosts` and that `/etc/hosts` are used by your system resolver.

@tailhook, any reason why we can't switch to `127.0.0.1` to make things work better in cases like this?

Are you able to attach your `dbschema/migrations` history?

The bug is likely path-dependent, i.e it's a combination of migrations that causes it. If I had to reduce this, I would take the innards of all migrations, concatenate them...

You can pipe into `edgedb`: ``` edgedb> CREATE DATABASE testdb; ``` ``` $ cat file.edgeql | edgedb -d testdb ```

No, the server should reload config from external files (e.g. certs) when it receives `SIGHUP`

One way would be to bind something like a `__conflict__` magic to the origin type of the constraint and prohibit binding the `insert` subject type in those cases. A bit...

(also, we _really_ need to be including the "detail" and "hint" in client binding error renderings)