Oliver, is this a meaningful test? I mean, on a new installed system where there is nothing do to... this sounds not like something I would like to see as a test for my enterprise Linux. What about waiting until the system is in some real life notebook state, like updates, sleep, updates, (fat updates, like kernel, ...., so that huge snapshots exist) , wait that something is to do, than do some disk intensive tasks, and than start the btrfs thing. than see if everything is still smooth, as on you newly installed system. automate this , have it as regression test, so that future patches btrfs fixes will not re trigger the test. (maybe I am naive, but this is what I would expect from something that calls itself enterprise linux)