Hide Forgot
Data layout of VMDK images with streamOptimized sub-format was incorrect Previously, the data layout of a Virtual Machine Disk (VMDK) image with a streamOptimized sub-format created by the qemu-img tool was incorrect. This prevented the VMDK image from being bootable when imported to ESX servers. In this update, the image is converted to a valid VMDK streamOptimized image. This results in the VMDK image being bootable.
Created attachment 1115668 [details] /tmp/example.ovf.sh Description of problem: Any VMDK image created by qemu-img is unbootable. Note that the import of the VMDK image via OVA fails already before (due to bug #1299116). Failure in VMware looks like this: --- snipp --- . error: unknown filesystem. Entering rescue mode... grub rescue> _ --- snapp --- Version-Release number of selected component (if applicable): qemu-img-1.5.3-105.el7_2.1.x86_64 VMware vSphere Client, Version 5.1.0, Build 1064113 VMware ESXi, Version 5.1.0, Build 1065491, German-000 How reproducible: Everytime, see above and below. Steps to Reproduce: 1. dd if=/dev/zero of=/tmp/example.img bs=1M count=1200 2. parted /tmp/example.img mktable msdos 3. parted /tmp/example.img mkpart primary ext4 1 -- -0 4. parted /tmp/example.img set 1 boot on 5. kpartx -avs /tmp/example.img 6. mkfs.ext4 /dev/mapper/loop0p1 7. mkdir -p /mnt/example/ 8. mount /dev/mapper/loop0p1 /mnt/example/ 9. yum -y --releasever=7 --nogpg --installroot=/mnt/example install \ systemd passwd centos-release kernel grub2 10. mount --bind /dev /mnt/example/dev 11. mount --bind /proc /mnt/example/proc 12. mount --bind /sys /mnt/example/sys 13. echo "GRUB_TIMEOUT=5" > /mnt/example/etc/default/grub 14. echo "GRUB_DISTRIBUTOR=\"\$(sed 's, release .*$,,g' \ /etc/system-release)\"" >> /mnt/example/etc/default/grub 15. echo "GRUB_DEFAULT=saved" >> /mnt/example/etc/default/grub 16. echo "GRUB_DISABLE_SUBMENU=true" >> /mnt/example/etc/default/grub 17. echo "GRUB_TERMINAL_OUTPUT=\"console\"" >> \ /mnt/example/etc/default/grub 18. echo "GRUB_CMDLINE_LINUX=\" vconsole.keymap=de \ vconsole.font=latarcyrheb-sun16\"" >> \ /mnt/example/etc/default/grub 19. echo "GRUB_DISABLE_RECOVERY=\"true\"" >> \ /mnt/example/etc/default/grub 20. chroot /mnt/example/ grub2-mkconfig -o /boot/grub2/grub.cfg 21. chroot /mnt/example/ grub2-install /dev/loop0 22. umount /mnt/example/dev 23. umount /mnt/example/proc 24. umount /mnt/example/sys 25. umount /mnt/example/ 26. kpartx -dvs /tmp/example.img 27. qemu-img convert /tmp/example.img -O vmdk \ -o adapter_type=lsilogic,subformat=streamOptimized,compat6 \ /tmp/example-disk1.vmdk 28. printf '\x03' | dd conv=notrunc of=/tmp/example-disk1.vmdk \ bs=1 seek=$((0x4)) 29. sh /tmp/example.ovf.sh 30. tar -C /tmp/ -cf /tmp/example.ova example.ovf example-disk1.vmdk 31. Import example.ova into VMware -> fails with error above Actual results: qemu-img created VMDK images are unbootable Expected results: qemu-img created VMDK images should be bootable. Repeating the steps above with qemu-img-2.5.0-3.fc24.x86_64 (rebuilt from Fedora Rawhide for RHEL 7) leads to success. Additional info: Virtual size alternatively from: qemu-img info /tmp/example-disk1.vmdk
Robert, thanks for the bugreport. As I said in the other BZ (bug 1299116), if this issue is critical or in any way time sensitive, please raise a ticket through your regular Red Hat support channels to make certain it receives the proper attention and prioritization that will result in a timely resolution. For information on how to contact the Red Hat production support team, please visit: https://www.redhat.com/support/process/production/#howto
Ademar, thanks for the reminder. Cross-filed case 01569297 on the Red Hat customer portal.
Does this mean there is something to test? If so, may you share the testing update with me (either here or via the ticket), please?
Yes, you can test the scratch build rpms that include the fixes for this bug (and bz 1299116): http://people.redhat.com/fzheng/bz1299250/
Thanks! Btw, is it intended that there is no SRPM? The %changelog does not mention anything, so I am wondering if it's really patched (before running my tests).
My reply to comment 7 from the other day got lost. Robert, the SRPM is not preserved but the packages do have the fix. Thanks, Fam
I would like to confirm that the updated RPM (qemu-img-1.5.3-108.el7.test) works fine for me and passed all my tests regarding this issue.
Great, thanks!
This bug was accidentally moved from POST to MODIFIED via an error in automation, please see mmccune@redhat.com with any questions
Fix included in qemu-kvm-1.5.3-111.el7
Reproduced the issue with qemu-kvm-1.5.3-110.el7.x86_64. After step 7, guest cannot boot up. Verified the issue with qemu-kvm-1.5.3-111.el7.x86_64(Step 3 is unnecessary). After step 7, guest can boot up successfully. Steps to reproduce: 1. Create a guest on ESXi 5.1 via vSphere and copy out the flat vmdk file(such as rhel6.8z_64_lazy-flat.vmdk, rename it as example.img. The format of the file is raw). 2. qemu-img convert example.img -O vmdk -o adapter_type=lsilogic,subformat=streamOptimized,compat6 example.vmdk 3. printf '\x03' | dd conv=notrunc of=example.vmdk bs=1 seek=$((0x4)) 4. sh example.ovf.sh 5. tar cf example.ova example.ovf example.vmdk 6. Import example.ova into ESXi 5.1. 7. Boot up the guest.
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory, and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://rhn.redhat.com/errata/RHSA-2016-2585.html