Comment 4 for bug 1503467

I checked the trigger functions themselves and they seem mostly OK - the only code that could possibly fire a node_update event for a device involves changing storage configuration of a node (which, if it's happening, would be a separate bug). Another hypothesis would be that those specific triggers are firing for the devices when they shouldn't, either by being misattached to the wrong events or by cascading update events. That's why we need to know what exactly happened and when, because something caused the device to show up on the wrong pipe.