I've dualcore x86_64 server with newest Ubuntu kernel 3.0.0-7-server. On this server is data storage builded from four (1.5T) discs in two raid1 arrays and joined with LVM to 3TB volume. This volume is formated with Btrfs a mounted with default options (rw,noatime). On server running corporate NFS and Samba server with same load every day. Everything works fine after reboot, but several days after server load (iowait cpu time) goes to high values and server performance is terrible. IOwait stays high at night too, when nobody works with this server. This situation is visible on munin graphs in attachments. After unmount and mount this storage again manually everything goes to normal for some next days. What I can try?
Created attachment 68332 [details] Munin graph of cpu load (2)
Created attachment 68342 [details] Munin graph of cpu load (1)
Similar simptoms are described at blog: http://blog.brendon.com/unix/are-we-in-with-the-new-btrfs-with-nfs-and-esxi/
Closing, if this is still affecting you on a newer kernel please reopen.