Kadischi error

[Greg DeKoenigsberg]

So I get this when trying to build the initrd:

===
making initrd image
/tmp/livecd-build_no12/system/lib/modules/None is not a directory.

*** Fatal error: /usr/local/share/kadischi/livecd-mkinitrd.sh returned 
non zero (256) exit code.  Aborting execution.
===

Any ideas?

--g

_____________________  ____________________________________________
  Greg DeKoenigsberg ] [ the future masters of technology will have
 Community Relations ] [ to be lighthearted and intelligent.  the
             Red Hat ] [ machine easily masters the grim and the 
                     ] [ dumb.  --mcluhan

[Skunk Worx]

Greg DeKoenigsberg wrote:
> So I get this when trying to build the initrd:
>
> ===
> making initrd image
> /tmp/livecd-build_no12/system/lib/modules/None is not a directory.

This is reported in the bugzilla. The only clue I have so far is 
attempting to run Kadischi on a yum (or otherwise) updated FC4 box 
appears to cause this problem.

I just installed box stock FC4 on an old K6 I have laying around and it 
created the iso fine. Will yum update the box tomorrow and see if it 
breaks Kadischi.

If it's not this it could be architecture or configuration related. Will 
post more when I know more...

--
SW

[Greg DeKoenigsberg]

On Wed, 12 Oct 2005, Skunk Worx wrote:
> Greg DeKoenigsberg wrote:
> > So I get this when trying to build the initrd:
> >
> > ===
> > making initrd image
> > /tmp/livecd-build_no12/system/lib/modules/None is not a directory.
>
> This is reported in the bugzilla. The only clue I have so far is 
> attempting to run Kadischi on a yum (or otherwise) updated FC4 box 
> appears to cause this problem.
>
> I just installed box stock FC4 on an old K6 I have laying around and it 
> created the iso fine. Will yum update the box tomorrow and see if it 
> breaks Kadischi.

Try just updating the kernel first.  At least we might be able to isolate 
the problem that way.

--g

_____________________  ____________________________________________
  Greg DeKoenigsberg ] [ the future masters of technology will have
 Community Relations ] [ to be lighthearted and intelligent.  the
             Red Hat ] [ machine easily masters the grim and the 
                     ] [ dumb.  --mcluhan

[Skunk Worx]

Greg DeKoenigsberg wrote:
> On Wed, 12 Oct 2005, Skunk Worx wrote:
> > Greg DeKoenigsberg wrote:
> > > So I get this when trying to build the initrd:
> > >
> > > ===
> > > making initrd image
> > > /tmp/livecd-build_no12/system/lib/modules/None is not a directory.
> >
> > This is reported in the bugzilla. The only clue I have so far is 
> > attempting to run Kadischi on a yum (or otherwise) updated FC4 box 
> > appears to cause this problem.
> >
> > I just installed box stock FC4 on an old K6 I have laying around and it 
> > created the iso fine. Will yum update the box tomorrow and see if it 
> > breaks Kadischi.
>
> Try just updating the kernel first.  At least we might be able to isolate 
> the problem that way.

still working through the latest kernel:

$ uname -a
Linux wombat 2.6.13-1.1526_FC4 #1 Wed Sep 28 19:15:03 EDT 2005 i586 i586 
i386 GNU/Linux

I have a local repo so I will probably go for a full update tomorrow 
night and see if it breaks.

If anyone wants me to try another set of pkgs. post a reply...

--
SW

[Darko Ilic]

On Thursday 13 October 2005 08:11, Skunk Worx wrote:
> > > > ===
> > > > making initrd image
> > > > /tmp/livecd-build_no12/system/lib/modules/None is not a directory.
> still working through the latest kernel:
>
> $ uname -a
> Linux wombat 2.6.13-1.1526_FC4 #1 Wed Sep 28 19:15:03 EDT 2005 i586 i586
> i386 GNU/Linux

Hmm, we are talking about updating the parent system, not the repository that 
will be used for live CD building, right?

Well, the kernel that is running on the parent system has nothing to do with 
the kernel on the live CD system.

I'm currently connecting to the internet through the GPRS so I can't download 
updates and try them by my self... But, there are two things you should pay 
attention to:

1. What repository are you using for live CD building? FC4 or not? 
2. Is the python on the parent system FC4 default, or is maybe updated?

Here's the function that is supposed to resolve the kernel version:

def get_kernel_version (rootdir):
 """Returns kernel version that is installed on the destination system"""
 kernel_version = None
 clear_rpm_db_files (rootdir)
 ts = rpm.TransactionSet (rootdir)
 for mi in ts.dbMatch ('name', 'kernel'):
  kernel_version = "%s-%s" % (mi['version'], mi['release'])
 clear_rpm_db_files (rootdir)
 return kernel_version

rootdir is the path to the root of the system. For the above example, it would 
be rootdir='/tmp/livecd-build_no12/system'. You could also try it out on your 
local system by calling get_kernel_version('/')

Just in case, here's the clear_rpm_db_files function that is used in 
get_kernel_version().

def clear_rpm_db_files (rootdir):
 """Removes rpm database files in the destination if they exist"""
 def rmpath (p):
  try: os.remove (p)
         except OSError: pass
 rmpath (normalize_path ('var/lib/rpm/__db.001', rootdir))
 rmpath (normalize_path ('var/lib/rpm/__db.002', rootdir))
 rmpath (normalize_path ('var/lib/rpm/__db.003', rootdir))

--
Darko

[Greg DeKoenigsberg]

On Thu, 13 Oct 2005, Darko Ilic wrote:
> On Thursday 13 October 2005 08:11, Skunk Worx wrote:
> > > > > ===
> > > > > making initrd image
> > > > > /tmp/livecd-build_no12/system/lib/modules/None is not a directory.
> > still working through the latest kernel:
> >
> > $ uname -a
> > Linux wombat 2.6.13-1.1526_FC4 #1 Wed Sep 28 19:15:03 EDT 2005 i586 i586
> > i386 GNU/Linux
>
> Hmm, we are talking about updating the parent system, not the repository that 
> will be used for live CD building, right?
>
> Well, the kernel that is running on the parent system has nothing to do with 
> the kernel on the live CD system.
>
> I'm currently connecting to the internet through the GPRS so I can't download 
> updates and try them by my self... But, there are two things you should pay 
> attention to:
> What repository are you using for live CD building? FC4 or not?
> Is the python on the parent system FC4 default, or is maybe updated?
>
> Here's the function that is supposed to resolve the kernel version:
>
> def get_kernel_version (rootdir):
>  """Returns kernel version that is installed on the destination system"""
>  kernel_version = None
>  clear_rpm_db_files (rootdir)
>  ts = rpm.TransactionSet (rootdir)
>  for mi in ts.dbMatch ('name', 'kernel'):
>   kernel_version = "%s-%s" % (mi['version'], mi['release'])
>  clear_rpm_db_files (rootdir)
>  return kernel_version

So it's returning "None" for me as kernel_version, which means it's 
falling through all of these cases.

I used stock FC4 as the repository.  I updated a number of packages on the
system itself via yum.  I tried running against two different kernels.

Note: I did build a new kernel on this box, though I never actually ran 
it.  Could building the kernel have changed some things that Kadischi 
assumes will be present by default?

And would it help if I gave you the contents of rpm.TransactionSet 
(rootdir) when I get this error?

--g

_____________________  ____________________________________________
  Greg DeKoenigsberg ] [ the future masters of technology will have
 Community Relations ] [ to be lighthearted and intelligent.  the
             Red Hat ] [ machine easily masters the grim and the 
                     ] [ dumb.  --mcluhan
> rootdir is the path to the root of the system. For the above example, it would 
> be rootdir='/tmp/livecd-build_no12/system'. You could also try it out on your 
> local system by calling get_kernel_version('/')
>
> Just in case, here's the clear_rpm_db_files function that is used in 
> get_kernel_version().
>
> def clear_rpm_db_files (rootdir):
>  """Removes rpm database files in the destination if they exist"""
>  def rmpath (p):
>   try: os.remove (p)
>          except OSError: pass
>  rmpath (normalize_path ('var/lib/rpm/__db.001', rootdir))
>  rmpath (normalize_path ('var/lib/rpm/__db.002', rootdir))
>  rmpath (normalize_path ('var/lib/rpm/__db.003', rootdir))
>
> --
> Darko
>
> --
> Fedora-livecd-list mailing list
> [e-mail]
> https://www.redhat.com/mailman/listinfo/fedora-livecd-list

[Darko Ilic]

On Thursday 13 October 2005 18:44, Greg DeKoenigsberg wrote:
> system itself via yum.  I tried running against two different kernels.
>
> Note: I did build a new kernel on this box, though I never actually ran
> it.  Could building the kernel have changed some things that Kadischi
> assumes will be present by default?

No, it shouldn't have anything to do with the kernel on the parent system.
> And would it help if I gave you the contents of rpm.TransactionSet
> (rootdir) when I get this error?

Hm, we can try, but I think ts object is just the interface to the rpm 
database.

I think it would help if you could edit /usr/share/kadischi/lib/functions.py, 
replace get_kernel_version  function with the following:

def get_kernel_version (rootdir):
  """Returns kernel version that is installed on the destination system"""
 kernel_version = None
 clear_rpm_db_files (rootdir)
 ts = rpm.TransactionSet (rootdir)
 mi = ts.dbMatch ()
 for hdr in mi:
  print "%s-%s-%s" % (hdr['name'], hdr['version'], hdr['release'])
 for mi in ts.dbMatch ('name', 'kernel'):
  kernel_version = "%s-%s" % (mi['version'], mi['release'])
 clear_rpm_db_files (rootdir)
 return kernel_version

And then send me the output. Kadischi will fail again, with the same error 
message, but it will print the list of packages that are present in the rpm 
database.

--
Darko

[Greg DeKoenigsberg]

Great.  I'll try to do that here in the next couple of hours.

--g

_____________________  ____________________________________________
  Greg DeKoenigsberg ] [ the future masters of technology will have
 Community Relations ] [ to be lighthearted and intelligent.  the
             Red Hat ] [ machine easily masters the grim and the 
                     ] [ dumb.  --mcluhan

On Thu, 13 Oct 2005, Darko Ilic wrote:
> On Thursday 13 October 2005 18:44, Greg DeKoenigsberg wrote:
> > system itself via yum.  I tried running against two different kernels.
> >
> > Note: I did build a new kernel on this box, though I never actually ran
> > it.  Could building the kernel have changed some things that Kadischi
> > assumes will be present by default?
>
> No, it shouldn't have anything to do with the kernel on the parent system.
> > And would it help if I gave you the contents of rpm.TransactionSet
> > (rootdir) when I get this error?
>
> Hm, we can try, but I think ts object is just the interface to the rpm 
> database.
>
> I think it would help if you could edit /usr/share/kadischi/lib/functions.py, 
> replace get_kernel_version  function with the following:
>
> def get_kernel_version (rootdir):
>   """Returns kernel version that is installed on the destination system"""
>  kernel_version = None
>  clear_rpm_db_files (rootdir)
>  ts = rpm.TransactionSet (rootdir)
>  mi = ts.dbMatch ()
>  for hdr in mi:
>   print "%s-%s-%s" % (hdr['name'], hdr['version'], hdr['release'])
>  for mi in ts.dbMatch ('name', 'kernel'):
>   kernel_version = "%s-%s" % (mi['version'], mi['release'])
>  clear_rpm_db_files (rootdir)
>  return kernel_version
>
> And then send me the output. Kadischi will fail again, with the same error 
> message, but it will print the list of packages that are present in the rpm 
> database.
>
> --
> Darko
>
> --
> Fedora-livecd-list mailing list
> [e-mail]
> https://www.redhat.com/mailman/listinfo/fedora-livecd-list

Replace isolinux with grub...

[Chyryll Zariss]

Is there anyway to replace isolinux with grub bootloader? It'll be 
pretty nice if it can boot with grub. (Just a suggestions).

Thanks..

Replace isolinux with grub...

[Ignacio Vazquez-Abrams]

On Fri, 2005-10-14 at 04:51 +0800, Chyryll Zariss wrote:
> Is there anyway to replace isolinux with grub bootloader? It'll be 
> pretty nice if it can boot with grub. (Just a suggestions).

http://www.gnu.org/software/grub/manual/html_node/Making-a-GRUB-bootable-CD_...

Replace isolinux with grub...

[Chyryll Zariss]

Roger that, I'll be following the grub howto..sound interesting..

After reading the howto, i notice especially the "root=/xxx" because in 
normal boot from harddisk usually use. Is it the same "/xxx" as in CD? I 
also put "rhgb quiet", does rhgb can be applied? How about the "kernel 
/boot/xx", "splashimage=/boot/xxx" and "initrd /boot/xxx", can be 
readable when booting from CD? One more thing, the root dev is ram0 right?

default=0
timeout=30
splashimage=/boot/splash.xpm.gz
title Fedora LiveCD
kernel /boot/vmlinuz-2.6.x.x root=/dev/ram0 rhgb quiet
initrd /boot/initrd-2.6.x.x

It's different when i look at linspire 5.0 menu.lst, the "/boot" have 
"(cd)" in the beginning:

splashimage=(cd)/boot/splash.xpm.gz
kernel (cd)/boot/vmlinuz-2.6.10
initrd (cd)/boot/initrd-2.6.10.gz

Many thanks..

Ignacio Vazquez-Abrams wrote:
> On Fri, 2005-10-14 at 04:51 +0800, Chyryll Zariss wrote:
> > Is there anyway to replace isolinux with grub bootloader? It'll be 
> > pretty nice if it can boot with grub. (Just a suggestions).
>
> http://www.gnu.org/software/grub/manual/html_node/Making-a-GRUB-bootable-CD_...
>
> --
> Fedora-livecd-list mailing list
> [e-mail]
> https://www.redhat.com/mailman/listinfo/fedora-livecd-list
>
> No virus found in this incoming message.
> Checked by AVG Anti-Virus.
> Version: 7.0.344 / Virus Database: 267.11.13/126 - Release Date: 09/10/2005

Replace isolinux with grub...

[Chyryll Zariss]

My way of adding grub failed when followed (as below) my previous 
posting. I found out the .sh file having an isolinux inside such as:

in install-boot.sh:
cp $sysdir/boot/isolinux/initrd.img $csysdir/boot/isolinux/initrd.img
cp $sysdir/boot/vmlinuz-$kernel $csysdir/boot/isolinux/vmlinuz
cp /usr/lib/syslinux/isolinux.bin $csysdir/boot/isolinux/
cat > $csysdir/boot/isolinux/isolinux.cfg <<_EOF_

in livecd-mkinitrd.sh:

imagedir="/boot/isolinux/"
image="initrd.img"

Anyhow to change this isolinux to grub folder/file  to read grub?

Thanks..

Chyryll Zariss wrote:
> Roger that, I'll be following the grub howto..sound interesting..
>
> After reading the howto, i notice especially the "root=/xxx" because 
> in normal boot from harddisk usually use. Is it the same "/xxx" as in 
> CD? I also put "rhgb quiet", does rhgb can be applied? How about the 
> "kernel /boot/xx", "splashimage=/boot/xxx" and "initrd /boot/xxx", can 
> be readable when booting from CD? One more thing, the root dev is ram0 
> right?
>
> default=0
> timeout=30
> splashimage=/boot/splash.xpm.gz
> title Fedora LiveCD
> kernel /boot/vmlinuz-2.6.x.x root=/dev/ram0 rhgb quiet
> initrd /boot/initrd-2.6.x.x
>
> It's different when i look at linspire 5.0 menu.lst, the "/boot" have 
> "(cd)" in the beginning:
>
> splashimage=(cd)/boot/splash.xpm.gz
> kernel (cd)/boot/vmlinuz-2.6.10
> initrd (cd)/boot/initrd-2.6.10.gz
>
> Many thanks..
>
> Ignacio Vazquez-Abrams wrote:
> > On Fri, 2005-10-14 at 04:51 +0800, Chyryll Zariss wrote:
> > > Is there anyway to replace isolinux with grub bootloader? It'll be 
> > > pretty nice if it can boot with grub. (Just a suggestions).
> >
> > http://www.gnu.org/software/grub/manual/html_node/Making-a-GRUB-bootable-CD_...
> >
> > -- 
> > Fedora-livecd-list mailing list
> > [e-mail]
> > https://www.redhat.com/mailman/listinfo/fedora-livecd-list
>
> -- 
> Fedora-livecd-list mailing list
> [e-mail]
> https://www.redhat.com/mailman/listinfo/fedora-livecd-list

[Skunk Worx]

the problem appears to be transient. the latest yum updates have fixed 
it. will retest tomorrow night to verify.
--
SW

[Skunk Worx]

Skunk Worx wrote:
> the problem appears to be transient. the latest yum updates have fixed 
> it. will retest tomorrow night to verify.
> -- 
> SW

I have verified that the problem :

/tmp/livecd-build_noNN/system/lib/modules/None is not a directory.

...is fixed.

It is not kernel related and appears to have been a transient problem in 
one of the updates packages.

If anyone see this problem on their kadischi build host and has a 
vanilla fc4 install (or has updated after Oct 10 16:30 PDT) please let 
me know.

--
SW

[Ignacio Vazquez-Abrams]

On Fri, 2005-10-14 at 14:23 -0700, Skunk Worx wrote:
> I have verified that the problem :
>
> /tmp/livecd-build_noNN/system/lib/modules/None is not a directory.
>
> ...is fixed.
>
> It is not kernel related and appears to have been a transient problem in 
> one of the updates packages.
>
> If anyone see this problem on their kadischi build host and has a 
> vanilla fc4 install (or has updated after Oct 10 16:30 PDT) please let 
> me know.

I'm still seeing it on my freshly-updated FC4 system. Somehow no kernel
is being installed into the buildroot.

[Skunk Worx]

Ignacio Vazquez-Abrams wrote:
> On Fri, 2005-10-14 at 14:23 -0700, Skunk Worx wrote:
> > If anyone see this problem on their kadischi build host and has a 
> > vanilla fc4 install (or has updated after Oct 10 16:30 PDT) please let 
> > me know.
>
> I'm still seeing it on my freshly-updated FC4 system. Somehow no kernel
> is being installed into the buildroot.

So you are getting the same msg :

/tmp/livecd-build_noNN/system/lib/modules/None is not a directory.

Thanks for the info, will re-test vanilla vs. latest updates tomorrow 
night on my hardware and post results.

--
SW

[Skunk Worx]

Skunk Worx wrote:
> Ignacio Vazquez-Abrams wrote:
> > On Fri, 2005-10-14 at 14:23 -0700, Skunk Worx wrote:
> > > If anyone see this problem on their kadischi build host and has a 
> > > vanilla fc4 install (or has updated after Oct 10 16:30 PDT) please 
> > > let me know.
> >
> > I'm still seeing it on my freshly-updated FC4 system. Somehow no kernel
> > is being installed into the buildroot.
>
> So you are getting the same msg :
>
> /tmp/livecd-build_noNN/system/lib/modules/None is not a directory.
>
> Thanks for the info, will re-test vanilla vs. latest updates tomorrow 
> night on my hardware and post results.
>
> -- 
> SW

i have tried it on three machines (both intel and amd mobos) and can't 
duplicate this problem. I'm guessing it's a deep problem below kadischi.

If anyone can provide more info on the chipset or cpu involved (N/S 
bridge etc./CPU) I can try to duplicate this bug.

--
SW