(In reply to Aaron Lefkowitz from comment #13) > Our workload has all kinds of extreme use cases in it. As Jeff Mahoney saw > when trying to debug why our workload was breaking btrfs performance which > led to an excellent bug fix in the allocator, the easiest way to reproduce > and debug our problems is to run in to them the same way we do. Because it's > just an openSUSE machine I assume you can get all kinds of hooks in before > seeing the failure. > > I'd be happy to show you our vagrant box which has this problem 100% of the > time and help you get a reproduce. It's not completely trivial to run, but > it is probably the easiest way for you to see what's going on. Let me know > how you'd like to proceed. Thanks for the proposal! Sounds good to me. I suggest scheduling a session in January?