From here: https://bugzilla.redhat.com/show_bug.cgi?id=1698088

When connected to multiple connections, if one conn is very slow it will delay UI stats reporting for all other connections, even if those connections are quick. This is because internally we use a single thread and queue for stats poll requests or object list update requests: a slow conn processing a request can back up the queue.

Also it's worth investigating how hard it is to have one thread per conn subsystem, so a thread for domains, pools, networks, interfaces, nodedevs, etc.