[Fedora-livecd-list] f7 livecd-creator cannot umount image after recent yum updates

[Skunk Worx]

All of our work is under strict CM so I can roll back the local yum repo 
  next week if I have to, then k/s the machines. Sorry about the top post.

---
John

Tim Wood wrote:
> FWIW and YMMV, but I started swearing by using a virtual machine (under 
> vmware) as my LiveCD development machine because of situations like 
> this.  I can't speak to Xen and qemu, but vmware does have a snapshot 
> feature.  I always snapshot a machine then run yum.  Then if an update 
> causes problems with iso creation, I go back to the snapshot.  Nothing 
> like having weird problems with iso creation hit when you're on a 
> deadline.  It's also the primary motivation behind my slowly plugging 
> away at a LiveCD creation LiveCD...  Can't get hosed if it doesn't 
> update...
> 
> Tim
> 
> 
> Skunk Worx wrote:
>> I ran yum updates two days ago and now the iso created by 
>> livecd-creator has a few issues. This is a Fedora 7 machine.
>>
>> 1) The tool complains that it cannot umount busy device :
>>
>> ...
>> Building an initramfs at /boot/livecd-initramfs.img for kernel 
>> 2.6.23.15-80.fc7
>> FATAL: Module ide_cd not found.
>> FATAL: Module usbhid not found.
>> Done; initramfs is 4.3M.
>>
>> umount: /var/tmp/livecd-creator-iiLVNk/install_root: device is busy
>> umount: /var/tmp/livecd-creator-iiLVNk/install_root: device is busy
>> ioctl: LOOP_CLR_FD: Device or resource busy
>> ...
>>
>> (otherwise completes successfully)
>>
>> 2) If I examine the final iso, mounting the squash image and then the 
>> os.img, I find the ext3 fs is inconsistent :
>>
>> mount -o loop /tmp/foo_sqsh/os.img /tmp/foo_img
>> EXT3-fs: write access unavailable, cannot proceed.
>> mount: wrong fs type, bad option, bad superblock on /dev/loop3,
>>        missing codepage or other error
>>        In some cases useful info is found in syslog - try
>>        dmesg | tail  or so
>>
>> dmesg :
>>
>> loop: module loaded
>> EXT3-fs: INFO: recovery required on readonly filesystem.
>> EXT3-fs: write access unavailable, cannot proceed.
>>
>> # file /tmp/foo_sqsh/os.img
>> /tmp/foo_sqsh/os.img: Linux rev 1.0 ext3 filesystem data (needs 
>> journal recovery) (large files)
>>
>> (If I copy the os.img out to a r/w location I can fsck.ext3 it)
>>
>> 3) The /etc/mtab in the os.img ends up errant so sysfs mount on boot 
>> of media complains :
>>
>> Remounting root filesystem in read-write mode:             [  OK  ]
>> Mounting local filesystems:  mount: sysfs already mounted or /sys busy
>> mount: according to mtab, /sys is already mounted on /sys
>>                                                            [FAILED]
>>
>> For now I am fixing the iso with a script :
>>
>> 1) copy out/fsck.ext3 the os.img file
>> 2) remove the errant /etc/mtab
>> 3) rebuild the squashfs with the new os.img using mksquashfs
>> 4) rebuild the iso
>>
>> I thought this was obviously a kernel problem, but reverting to an 
>> older  kernel on the build host did not fix it.
>>
>> TIA,
>> John
>>
>> -- 
>> Fedora-livecd-list mailing list
>> Fedora-livecd-list at redhat.com
>> https://www.redhat.com/mailman/listinfo/fedora-livecd-list
>>
> 
>