Hey  all, i've recenty upgraded my diskless nfsroot systems from using
   aufs1/2.6.27 to aufs2/2.6.31 (and 2.6.32).  I'm having an issue with aufs2
   that seems..well..strange.
   System details:
   =============
   Netbooted diskless system, uses a readonly nfsroot and a rw tmpfs.  System
   root is setup via a custom init script (makes the tmpfs, overlays the tmpfs
   ontop of the root fs with aufs, calls pivot_root ..pretty standard diskless
   stuff), uses:
       mktmpfs /mnt/ramdisk
       mount -n -t aufs -o br:/mnt/ramdisk=rw:/=ro none /mnt/initrd
   ..as the aufs mount command.  I'm using the gentoo package of aufs2-20100514
   , which is based on aufs2-standalone iirc.  i'm compiling with inotify
   support  but  not with nfs-export support (i don't export nfs from the
   diskless systems) and the kernel is being patched appropriately.  dmesg
   shows no errors or messages..
   Important notes after bootup:
   - original nfs mount for / is in /mnt/initrd/
   - original tmpfs mount is in /mnt/initrd/mnt/ramdisk/
   The Problem:
   ============
   A  few various directories within the filesystem are being erroneously
   copied-up(?? i don't know the word for this) to the tmpfs for no apparent
   reason  and are virtually empty in comparison with the contents of the
   nfsroot.   Which  directories  these are seem to change every reboot..
   Example:
   Usually, if I say, want to try and recompile a kernel -- everything's fine
   to begin with.
   cd /usr/src/linux ; make oldconfig
   After that completes, /mnt/initrd/mnt/ramdisk/usr/src/include/linux may
   (usually but not always) exists and is usually empty or nearly empty.  The
   parent  dir  only  holds a subdir 'linux'.  This effectively nullifies
   /usr/src/linux/include/linux which makes the next stage of compiling a
   kernel rather difficult.
   I can, however, rm -f /mnt/initrd/mnt/ramdisk/usr/src/include  ..and then
   everything's back to normal.
   So, since I have no idea where to even start, any suggestions on mount
   options to try or other methods to debug this?
   Thanks,
   Ian
------------------------------------------------------------------------------
ThinkGeek and WIRED's GeekDad team up for the Ultimate 
GeekDad Father's Day Giveaway. ONE MASSIVE PRIZE to the 
lucky parental unit.  See the prize list and enter to win: 
http://p.sf.net/sfu/thinkgeek-promo

Reply via email to