headers
sfjro | 1 Feb 2010 07:34
Picon

my local repository broke again


Today's release is skipped since my my local repository broke again.
The release would contain one bugfix.

- aufs: bugfix, ioctl(2) returns ENOTTY instead of EINVAL
  Reported by Stew Benedict.

I am not sure when I can fix my repository. If it succeeds at once, I
will release the latest aufs2 tommorrow.

J. R. Okajima

------------------------------------------------------------------------------
The Planet: dedicated and managed hosting, cloud storage, colocation
Stay online with enterprise data centers and the best network in the business
Choose flexible plans and management services without long-term contracts
Personal 24x7 support from experience hosting pros just a phone call away.
http://p.sf.net/sfu/theplanet-com
Permalink | Reply | 3 comments |
headers
sfjro | 1 Feb 2010 10:03
Picon

Re: my local repository broke again


> Today's release is skipped since my my local repository broke again.
> The release would contain one bugfix.
> 
> - aufs: bugfix, ioctl(2) returns ENOTTY instead of EINVAL
>   Reported by Stew Benedict.
> 
> I am not sure when I can fix my repository. If it succeeds at once, I
> will release the latest aufs2 tommorrow.

Anyone who is familiar with GIT, please help me.
Today, I tried git push as usual, but got errors.

error: packfile ./objects/pack/pack-f3972e40fa541573e9388780100d3fa4b4bb65a2.pack does not
match index
error: packfile ./objects/pack/pack-f3972e40fa541573e9388780100d3fa4b4bb65a2.pack cannot be accessed
error: refs/heads/aufs2 does not point to a valid object!
error: packfile ./objects/pack/pack-f3972e40fa541573e9388780100d3fa4b4bb65a2.pack does not
match index
error: packfile ./objects/pack/pack-f3972e40fa541573e9388780100d3fa4b4bb65a2.pack cannot be accessed
error: refs/heads/aufs2-27 does not point to a valid object!
error: packfile ./objects/pack/pack-f3972e40fa541573e9388780100d3fa4b4bb65a2.pack does not
match index
	:::

There were several lines saying "refs/heads/≤branch_name> does not point
to a valid object!" for every branch I tried pushing.

git status, checkout were fine.
Then I did git gc. It exited silently. And then git fsck. It reported
(Continue reading)

Permalink | Reply | 1 comment |
headers
Hans-Peter Jansen | 1 Feb 2010 15:38

Re: my local repository broke again

On Monday 01 February 2010, 10:03:54 sfjro <at> users.sourceforge.net wrote:
> > Today's release is skipped since my my local repository broke again.
> > The release would contain one bugfix.
> >
> > - aufs: bugfix, ioctl(2) returns ENOTTY instead of EINVAL
> >   Reported by Stew Benedict.
> >
> > I am not sure when I can fix my repository. If it succeeds at once, I
> > will release the latest aufs2 tommorrow.
>
> Anyone who is familiar with GIT, please help me.
> Today, I tried git push as usual, but got errors.
>
> error: packfile
> ./objects/pack/pack-f3972e40fa541573e9388780100d3fa4b4bb65a2.pack does
> not match index error: packfile
> ./objects/pack/pack-f3972e40fa541573e9388780100d3fa4b4bb65a2.pack cannot
> be accessed error: refs/heads/aufs2 does not point to a valid object!
> error: packfile
> ./objects/pack/pack-f3972e40fa541573e9388780100d3fa4b4bb65a2.pack does
> not match index error: packfile
> ./objects/pack/pack-f3972e40fa541573e9388780100d3fa4b4bb65a2.pack cannot
> be accessed error: refs/heads/aufs2-27 does not point to a valid object!
> error: packfile
> ./objects/pack/pack-f3972e40fa541573e9388780100d3fa4b4bb65a2.pack does
> not match index
>
>
> There were several lines saying "refs/heads/≤branch_name> does not point
> to a valid object!" for every branch I tried pushing.
(Continue reading)

Permalink | Reply | 0 comments |
headers
Hans-Peter Jansen | 1 Feb 2010 17:26

Re: aufs2 on nfs3 root failing

Dear Junjiro,

On Thursday 14 January 2010, 04:27:14 sfjro <at> users.sourceforge.net wrote:
> Hi Hans,
>
> "Hans-Peter Jansen":
> > for aufs2, I tried these mount variants:
> >
> > 	mount -t aufs -o br:/rwroot=rw:/roroot=ro none /root
> > resulted in:
> > 	aufs au_xino_create:659:mount[206]: xino doesn't support
> > /tmp/.aufs.xino(rootfs)
> >
> > 	mount -t aufs -o br:/rwroot=rw:/roroot=ro,xino=/rwroot/.aufs.xino none
> > /root resulted in:
> > 	aufs au_xino_create:659:mount[206]: xino doesn't support
> > /rwroot/.aufs.xino(nfs)
> >
> > 	mount -t aufs -o br:/rwroot=rw:/roroot=ro,noxino none /root
> > resulted in:
> > 	aufs au_opts_mount:1456:mount[206]: no branches
> >
> > So far, I'm unable to use aufs2 in my setup, which is a real pity.
> >
> > Given, that tmpfs seems to be unable to carry the xino file, and so
> > does nfs, how is a NFS3 diskless setup supposed to work now?
> >
> > Do you have any ideas or recommendations on how to precede?
>
> Some filesystems doesn't have the neccessary features for aufs XINO
(Continue reading)

Permalink | Reply | 1 comment |
headers
sfjro | 1 Feb 2010 18:09
Picon

Re: aufs2 on nfs3 root failing


Hi,

"Hans-Peter Jansen":
> Okay, here we go. I'm preparing the client setup with kiwi now, and based 
> it on openSUSE 11.1. Since I try to use it out of the box (SuSE still 
> carries a heavily patched kernel..), I'm down to:
> 
>  - kernel 2.6.27.42
>  - aufs 20081020
> 
> Yes, still a overly old aufs in the equation (sorry about that), but one, 
> that is strongly exercised (at least with local filesystems). I adjusted 

I am forgetting about aufs1, but if I remember correctly, aufs1 needed
CONFIG_AUFS_brabra and a patch to support nfs branch.

> the kiwi initrd in a way, that two nfs3 mounts are unioned, where the xino 
> mapping file resides on tmpfs. The problem is: during initrd, all is well, 
> but after transition to the real root, the aufs / fs is magically turned ro
> behind the scenes. The same happens, if the rw part is located on tmpfs as
> well.

You might need to move /read-write and /xino (probably) under the new
root dir, if you use initramfs.

> Does this rings a bell for you?

Aufs1 will not ring my bell any more. :-)
(Continue reading)

Permalink | Reply | 0 comments |
headers
Lucas de Vries | 1 Feb 2010 23:38
Gravatar

Stale NFS file handle

Hello,

I've been trying to search for a suitable union filesystem for a while,
but all of them are either buggy or buggy *and* slow. I decided to try
aufs, and so far all the features seem great, there's just one big bug
I just encountered.

I tried to compile something in my aufs directory (I've tried it with
multiple applications, so it's not something specific to that), and
during the configure stage it just dies with:

ls: cannot access .: Stale NFS file handle

I'm using AUFS compiled as a module. Applied the -standalone and -base
patches to my kernel as the README said. I'm not sure what I
could do now to fix this, hence my mail here.

I'm mounting three btrfs partitions with the following fstab line:

data /data aufs br:/mnt/data-5=rw:/mnt/data-4=rw:/mnt/data-3=rw,create=pmfs,noatime,noauto 0 0

The noauto is there because my data partitions only get decrypted
through LUKS at a later stage during the boot process, and I manually
mount it with "mount /data" after they are mounted.

I have not been able to reproduce the error with manual ls commands, it
only ever occurs during compilation.

Any help would be appreciated, I'm getting rather desparate in my search
for a non-suck unioning filesystem.
(Continue reading)

Permalink | Reply | 3 comments |
headers
sfjro | 2 Feb 2010 02:41
Picon

Re: Stale NFS file handle


Hello Lucas,

Lucas de Vries:
> I tried to compile something in my aufs directory (I've tried it with
> multiple applications, so it's not something specific to that), and
> during the configure stage it just dies with:
> 
> ls: cannot access .: Stale NFS file handle

Give me these info please.

(from the aufs README)
----------------------------------------------------------------------
- /proc/mounts (instead of the output of mount(8))
- /sys/module/aufs/*
- /sys/fs/aufs/* (if you have them)
- /debug/aufs/* (if you have them)
- linux kernel version
  if your kernel is not plain, for example modified by distributor,
  the url where i can download its source is necessary too.
- aufs version which was printed at loading the module or booting the
  system, instead of the date you downloaded.
- configuration (define/undefine CONFIG_AUFS_xxx)
- kernel configuration or /proc/config.gz (if you have it)
- behaviour which you think to be incorrect
- actual operation, reproducible one is better
- mailto: aufs-users at lists.sourceforge.net
----------------------------------------------------------------------
(Continue reading)

Permalink | Reply | 0 comments |
headers
sfjro | 2 Feb 2010 03:57
Picon

aufs2 GIT release (Re: my local repository broke again)


> Today's release is skipped since my my local repository broke again.
	:::
> I am not sure when I can fix my repository. If it succeeds at once, I
> will release the latest aufs2 tommorrow.

Bruno created a new repository and replaced aufs2-2.6.git.
Thank you very much.
I will upgrade my GIT tool and run git gc every week, I am not sure it
can stop the problem though.

Here is the latest aufs2 GIT release.
Users can git pull now.

o bugfix
- aufs: bugfix, ioctl(2) returns ENOTTY instead of EINVAL
  Reported by Stew Benedict.

J. R. Okajima

----------------------------------------------------------------------

- aufs2-2.6.git
      aufs: bugfix, ioctl(2) returns ENOTTY instead of EINVAL
      aufs: cosmetic clean-up for inode.c
      aufs: cosmetic clean-up for dentry.c

- aufs2-standalone.git
  ditto
(Continue reading)

Permalink | Reply | 0 comments |
headers
Wilkinson, Alex | 2 Feb 2010 04:17
Picon

Re: masking out subdirectories ... is it possible ? [SEC=UNCLASSIFIED]


    0n Fri, Jan 29, 2010 at 10:23:56AM -0600, Michael S. Zick wrote: 

    >> (generic)
    >> $ mkdir /tmp/empty
    >> # mount -o bind /tmp/empty /mnt/aufs/you/want/to/hide

This works great! However, it doesnt look like i can add this to fstab(5).
How would one normally get this automagically mounted ? rc.local ?
Is it possible to use AutoFS to automagically do a --bind mount ?

distro: debian.

Thanks!

   -Alex

IMPORTANT: This email remains the property of the Australian Defence Organisation and is subject to the
jurisdiction of section 70 of the CRIMES ACT 1914.  If you have received this email in error, you are
requested to contact the sender and delete the email.

------------------------------------------------------------------------------
The Planet: dedicated and managed hosting, cloud storage, colocation
Stay online with enterprise data centers and the best network in the business
Choose flexible plans and management services without long-term contracts
Personal 24x7 support from experience hosting pros just a phone call away.
http://p.sf.net/sfu/theplanet-com
Permalink | Reply | 5 comments |
headers
sfjro | 2 Feb 2010 04:34
Picon

Re: masking out subdirectories ... is it possible ? [SEC=UNCLASSIFIED]


"Wilkinson, Alex":
>     >> (generic)
>     >> $ mkdir /tmp/empty
>     >> # mount -o bind /tmp/empty /mnt/aufs/you/want/to/hide
> 
> This works great! However, it doesnt look like i can add this to fstab(5).
> How would one normally get this automagically mounted ? rc.local ?
> Is it possible to use AutoFS to automagically do a --bind mount ?

If you put the aufs entry in your fstab, try writing

/empty /mnt/aufs/you/want/to/hide none bind 0 0

after the aufs line.
Of course, you need to create /empty dir before you reboot.

But if your mount(8) behaves asynchronously, it may not work.

J. R. Okajima

------------------------------------------------------------------------------
The Planet: dedicated and managed hosting, cloud storage, colocation
Stay online with enterprise data centers and the best network in the business
Choose flexible plans and management services without long-term contracts
Personal 24x7 support from experience hosting pros just a phone call away.
http://p.sf.net/sfu/theplanet-com
Permalink | Reply | 3 comments |

Gmane