Comment 2 for bug 1420057

Took a look at this one this afternoon. I tried to get the API client to leak sockets in a simple standalone main, to no avail. I'm not convinced it was the API client connections that leaked fds here.

Seems suspicious to me that there were a large number of prior mongo errors that got us to the point of exhausting fds though.

What was the mongo configuration on this state server like? Was it clustered?