Eibo Thieme:
> > - Is the root dir ok? In other words, is NFS ok?
:::
> If I am not mistaken, NFS is ok. Everything on the root fs stays=20
Ok.
> (actually I am wondering about those old whiteout files, must be=20
> leftovers of former installations)
Ok, they should never do harm.
> > - Is the mount tree traversal ok?
:::
> > - Is procfs ok?
> > This is just-to-make-sure point too. Try "stat /proc; stat -f /proc"=
> .
:::
> stat /proc before error
:::
> stat -f /proc before error
Ok, these outputs are totally correct.
> After error:
>
> ls -la /proc after error
> =3D=3D=3D
> insgesamt 8
> dr-xr-xr-x 2 root root 4096 24. M=C3=A4r 2012 .
> drwxr-xr-x 48 root root 4096 22. Feb 11:01 ..
>
> stat /proc after error
> =3D=3D=3D
> File: =E2=80=9E/proc=E2=80=9C
> Size: 4096 Blocks: 8 IO Block: 32768 Verzeichnis
> Device: 10h/16d Inode: 17 Links: 2
> Access: (0555/dr-xr-xr-x) Uid: ( 0/ root) Gid: ( 0/ root)
> Access: 2013-02-21 15:41:15.000000000 +0100
> Modify: 2012-03-24 13:01:37.000000000 +0100
> Change: 2012-03-24 14:48:40.000000000 +0100
>
> stat -f /proc after error
> =3D=3D=3D
> File: "/proc"
> ID: 0 Namelen: 242 Type: UNKNOWN (0x61756673)
> Block size: 32768 Fundamental block size: 32768
> Blocks: Total: 600881 Free: 394835 Available: 364312
> Inodes: Total: 1221600 Free: 903632
These outputs are totally wrong.
I am afraid your "/bin/mount --move /proc ${TARGET}/proc" doesn't work,
or if it really works then traversing mount tree is broken.
I'd suggest you to try these instead of "mount --move"
- mount --bind /proc ${TARGET}/proc
or
- mount -t proc ${TARGET}/proc
> Actually, I use two aufs mounts, the root mount is the setup you have=20
> seen with private on top of general client on top of server root. This=20
> gives me an amazingly simple way to administrate these machines, I can=20
> install anything on the server and it just works on the clients. Adding=20
> a client is a trivial. Over the years I have come to the conclusion,=20
> that to keep it simple, I have to use aufs for /var as well, reducing=20
Actually this usage is my ideal use-case of aufs. In other words, I
want a good solution for such case. (I have written and released a
sample script to do this in aufs1 era).
> There is one phenonmenon here, as both mounts /aufs/server-var and=20
> /aufs/client-var are still there, even after mounting. It might be that=20
> only mounts on mountpoints in the root directory are affected. I will=20
> try to check this out in more detail.
In your /proc/mounts, I can see /var/run and /var/lock are mounted
before aufs-/var. I am afraid these tmpfs are hidden?
J. R. Okajima
------------------------------------------------------------------------------
Everyone hates slow websites. So do we.
Make your web apps faster with AppDynamics
Download AppDynamics Lite for free today:
http://p.sf.net/sfu/appdyn_d2d_feb
