Yesterday after the intervention I poked around for a bit and ended up finding a possible cause of the server side memory leak, fixed that, and overnight the memory use has been flat so I’m happy to conclude that the work has been upgraded from an intervention to a solution.