I noticed a webgrid node with a load average over 300 today, it seems that no matter where it gets scheduled, osm4wiki is running enough simultaneous perl processes that it takes over the node, basically.
Restarting the process didn't help. It seems to hang when I try to access to tool via the web, so I think it is malfunctioning somehow.
I did migrate the osmdb service from labsdb1006/7 to cloud internal services, but restarting the service didn't seem to help. I still cannot get a response from it, but it generates lots of load.
Please take a look and let me know if I can help.