Hi everyone, This year, I'm hoping to discuss interactions between filesystems, the MM code, the IO layer and resource controls. I'll come armed with data from the fleet and I'm especially interested in finding ways to test future changes to make sure resource isolation features don't regress. Other related topics are offloading CPU heavy work (like compression) and still doing the IO from the right cgroup. I'm reworking the btrfs compression flow right now because it ends up creating priority inversions. Mostly, I'm looking forward to syncing with everyone and catching up on development plans. -chris
