5 Replies Latest reply on Feb 26, 2011 2:39 AM by goldmann

    CentOS on CentOS image build issues

    tdyas

      I installed BoxGrinder on my CentOS 5.5 workstation. I have encountered some problems in building a CentOS VM image in this environment.  I solved or worked around some of the problems, but others remain.  Here are my notes:

       

      1.  Appliance-creator was throwing errors during the RPM install process related to "lsetfilecon."  Putting SELinux on the host into Permissive mode did not work.  I had to disable SELinux entirely on the host.

       

      2.  The version of parted that comes with CentOS is buggy with respect to requesting the kernel to re-read a parition table on a loop device.  I had to upgrade parted to the latest upstream GNU release (2.3) in order to get it to work properly. (I forward ported the patches in the CentOS RPM that had not been upstreamed and remade the parted RPM if anyone wants to take a look at it. There is a problem with .so versioning with the RPM but that can probably be solved easily enough.)

       

      3.  I believe I've encountered a race condition in the install process where the RPM installation is still proceeding when the code attempts to run the grub install.  I encountered a version of this race condition when I was building a Fedora VM on a Fedora host.  I tracked it to a RPM package with non-trivial %post scripts that was not completing instantly.  I removed that package from the appliance and the boxgrinder failure went away.

       

      Here is relevant snippet of my BoxGrinder log:

       

       

      D, [2011-02-23T18:03:33.234064 #4472] DEBUG -- : Installing: pm-utils                     ##############

      ####### [290/291]

      D, [2011-02-23T18:03:34.298630 #4472] DEBUG -- : Writing grub /home/tdyas/BoxGrinder/build/appliances/x8

      6_64/centos/5/centos-node/centos-plugin/tmp/imgcreate-mDrAaF/install_root/boot/grub/device.map

      D, [2011-02-23T18:03:36.853613 #4472] DEBUG -- : ls: /home/tdyas/BoxGrinder/build/appliances/x86_64/cent

      os/5/centos-node/centos-plugin/tmp/imgcreate-mDrAaF/install_root/boot/initramfs*: No such file o

      r directory

      D, [2011-02-23T18:03:36.856012 #4472] DEBUG -- : Writing grub config /home/tdyas/BoxGrinder/build/applia

      nces/x86_64/centos/5/centos-node/centos-plugin/tmp/imgcreate-mDrAaF/install_root/boot/grub/grub.

      conf

      D, [2011-02-23T18:03:36.856191 #4472] DEBUG -- : Copying /home/tdyas/BoxGrinder/build/appliances/x86_64/

      centos/5/centos-node/centos-plugin/tmp/imgcreate-mDrAaF/install_root/usr/share/grub/x86_64-redha

      t/e2fs_stage1_5 to /home/tdyas/BoxGrinder/build/appliances/x86_64/centos/5/centos-node/centos-pl

      ugin/tmp/imgcreate-mDrAaF/install_root/boot/grub/e2fs_stage1_5

      D, [2011-02-23T18:03:36.868901 #4472] DEBUG -- : Copying /home/tdyas/BoxGrinder/build/appliances/x86_64/

      centos/5/centos-node/centos-plugin/tmp/imgcreate-mDrAaF/install_root/usr/share/grub/x86_64-redha

      t/stage1 to /home/tdyas/BoxGrinder/build/appliances/x86_64/centos/5/centos-node/centos-plugin/tm

      p/imgcreate-mDrAaF/install_root/boot/grub/stage1

      D, [2011-02-23T18:03:36.870879 #4472] DEBUG -- : Copying /home/tdyas/BoxGrinder/build/appliances/x86_64/

      centos/5/centos-node/centos-plugin/tmp/imgcreate-mDrAaF/install_root/usr/share/grub/x86_64-redha

      t/stage2 to /home/tdyas/BoxGrinder/build/appliances/x86_64/centos/5/centos-node/centos-plugin/tm

      p/imgcreate-mDrAaF/install_root/boot/grub/stage2

      D, [2011-02-23T18:03:38.757879 #4472] DEBUG -- : Installing grub to /dev/loop1

      D, [2011-02-23T18:03:38.845241 #4472] DEBUG -- : Probing devices to guess BIOS drives. This may take a l

      ong time.

      D, [2011-02-23T18:03:38.926735 #4472] DEBUG -- : Installing: kudzu                        ##############

      ####### [291/291]

      D, [2011-02-23T18:03:38.926846 #4472] DEBUG -- :

      D, [2011-02-23T18:03:38.926904 #4472] DEBUG -- : GNU GRUB  version 0.97  (640K lower / 3072K upper memor

      y)

      D, [2011-02-23T18:03:38.926954 #4472] DEBUG -- :

      D, [2011-02-23T18:03:38.927005 #4472] DEBUG -- : [ Minimal BASH-like line editing is supported.  For the

      first word, TAB

      D, [2011-02-23T18:03:38.927062 #4472] DEBUG -- : lists possible command completions.  Anywhere else TAB

      lists the possible

      D, [2011-02-23T18:03:38.927116 #4472] DEBUG -- : completions of a device/filename.]

      D, [2011-02-23T18:03:38.927184 #4472] DEBUG -- : grub> device (hd0) /dev/loop1

      D, [2011-02-23T18:03:38.927241 #4472] DEBUG -- : grub> root (hd0,0)

       

      You will note that kudzu installed after the grub stage had already started. Is there any way to make appliance-creator wait for the RPM install stage to complete?

       

      -Tom

        • 1. CentOS on CentOS image build issues
          tdyas

          Forgot to include the actual error that killed the build.  Basically, BoxGrinder cannot unmount the loop device and then cannot find the raw disk image.  Here is the remainder of the log file:

           

          D, [2011-02-23T18:03:38.927302 #4472] DEBUG -- : Filesystem type is ext2fs, partition type 0x83

          D, [2011-02-23T18:03:38.927366 #4472] DEBUG -- : grub> setup --stage2=/home/tdyas/BoxGrinder/build/appli

          ances/x86_64/centos/5/centos-node/centos-plugin/tmp/imgcreate-mDrAaF/install_root/boot/grub/stag

          e2 --prefix=/grub  (hd0)

          D, [2011-02-23T18:03:38.927423 #4472] DEBUG -- : Checking if "/grub/stage1" exists... yes

          D, [2011-02-23T18:03:38.927476 #4472] DEBUG -- : Checking if "/grub/stage2" exists... yes

          D, [2011-02-23T18:03:38.927535 #4472] DEBUG -- : Checking if "/grub/e2fs_stage1_5" exists... yes

          D, [2011-02-23T18:03:38.927594 #4472] DEBUG -- : Running "embed /grub/e2fs_stage1_5 (hd0)"... failed (th

          is is not fatal)

          D, [2011-02-23T18:03:38.927741 #4472] DEBUG -- : Running "embed /grub/e2fs_stage1_5 (hd0,0)"... failed (

          this is not fatal)

          D, [2011-02-23T18:03:38.927835 #4472] DEBUG -- : Running "install --stage2=/home/tdyas/BoxGrinder/build/

          appliances/x86_64/centos/5/centos-node/centos-plugin/tmp/imgcreate-mDrAaF/install_root/boot/grub

          /stage2 /grub/stage1 (hd0) /grub/stage2 p /grub/grub.conf "... succeeded

          D, [2011-02-23T18:03:38.927924 #4472] DEBUG -- : Done.

          D, [2011-02-23T18:03:38.928010 #4472] DEBUG -- : grub> quit

          D, [2011-02-23T18:03:43.410515 #4472] DEBUG -- : /sbin/setfiles:  labeling files under /

          D, [2011-02-23T18:03:49.873074 #4472] DEBUG -- : /sbin/setfiles:  labeling files under /

          D, [2011-02-23T18:03:49.873287 #4472] DEBUG -- : matchpathcon_filespec_eval:  hash table stats: 33730 el

          ements, 23879/65536 buckets used, longest chain length 3

          D, [2011-02-23T18:03:49.873579 #4472] DEBUG -- : /sbin/setfiles:  Done.

          D, [2011-02-23T18:03:49.970498 #4472] DEBUG -- : umount: /home/tdyas/BoxGrinder/build/appliances/x86_64/

          centos/5/centos-node/centos-plugin/tmp/imgcreate-mDrAaF/install_root/proc: device is busy

          D, [2011-02-23T18:03:49.972629 #4472] DEBUG -- : umount: /home/tdyas/BoxGrinder/build/appliances/x86_64/

          centos/5/centos-node/centos-plugin/tmp/imgcreate-mDrAaF/install_root/proc: device is busy

          D, [2011-02-23T18:03:49.975505 #4472] DEBUG -- : Unmounting directory /home/tdyas/BoxGrinder/build/appli

          ances/x86_64/centos/5/centos-node/centos-plugin/tmp/imgcreate-mDrAaF/install_root/boot

          D, [2011-02-23T18:03:49.981355 #4472] DEBUG -- : Unmounting directory /home/tdyas/BoxGrinder/build/appli

          ances/x86_64/centos/5/centos-node/centos-plugin/tmp/imgcreate-mDrAaF/install_root/

          D, [2011-02-23T18:03:49.984921 #4472] DEBUG -- : umount: /home/tdyas/BoxGrinder/build/appliances/x86_64/

          centos/5/centos-node/centos-plugin/tmp/imgcreate-mDrAaF/install_root: device is busy

          D, [2011-02-23T18:03:49.985013 #4472] DEBUG -- : umount: /home/tdyas/BoxGrinder/build/appliances/x86_64/

          centos/5/centos-node/centos-plugin/tmp/imgcreate-mDrAaF/install_root: device is busy

          D, [2011-02-23T18:03:49.985359 #4472] DEBUG -- : Removing compat symlinks

          D, [2011-02-23T18:03:49.985546 #4472] DEBUG -- : Unmapping /dev/loop1

          D, [2011-02-23T18:03:50.613422 #4472] DEBUG -- : device-mapper: remove ioctl failed: Device or resource busy

          D, [2011-02-23T18:03:50.614636 #4472] DEBUG -- : Losetup remove /dev/loop1

          D, [2011-02-23T18:03:50.650817 #4472] DEBUG -- : ioctl: LOOP_CLR_FD: Device or resource busy

          D, [2011-02-23T18:03:50.652266 #4472] DEBUG -- : moving disks to stage location

          D, [2011-02-23T18:03:50.652916 #4472] DEBUG -- : moving /home/tdyas/BoxGrinder/build/appliances/x86_64/centos/5/centos-node/centos-plugin/tmp/imgcreate-mDrAaF/tmp-aFeAxA/centos-node-sda.raw to /home/tdyas/BoxGrinder/build/appliances/x86_64/centos/5/centos-node/centos-plugin/tmp/imgcreate-mDrAaF/out/centos-node-sda.raw

          D, [2011-02-23T18:03:50.744575 #4472] DEBUG -- : writing image XML to /home/tdyas/BoxGrinder/build/appliances/x86_64/centos/5/centos-node/centos-plugin/tmp/imgcreate-mDrAaF/out/centos-node.xml

          D, [2011-02-23T18:03:50.745108 #4472] DEBUG -- : creating destination dir: build/appliances/x86_64/centos/5/centos-node/centos-plugin/tmp/centos-node

          D, [2011-02-23T18:03:50.801805 #4472] DEBUG -- : moving /home/tdyas/BoxGrinder/build/appliances/x86_64/centos/5/centos-node/centos-plugin/tmp/imgcreate-mDrAaF/out/centos-node-sda.raw to build/appliances/x86_64/centos/5/centos-node/centos-plugin/tmp/centos-node/centos-node-sda.raw

          D, [2011-02-23T18:03:50.802165 #4472] DEBUG -- : moving /home/tdyas/BoxGrinder/build/appliances/x86_64/centos/5/centos-node/centos-plugin/tmp/imgcreate-mDrAaF/out/centos-node.xml to build/appliances/x86_64/centos/5/centos-node/centos-plugin/tmp/centos-node/centos-node.xml

          D, [2011-02-23T18:03:57.967059 #4472] DEBUG -- :

          D, [2011-02-23T18:03:57.967285 #4472] DEBUG -- :

          D, [2011-02-23T18:03:57.967416 #4472] DEBUG -- : Finished

          D, [2011-02-23T18:03:58.006750 #4472] DEBUG -- : Preparing guestfs...

          D, [2011-02-23T18:03:58.170120 #4472] DEBUG -- : Launching guestfs...

          F, [2011-02-23T18:04:59.488574 #4472] FATAL -- : No root partition found for 'centos-node-sda.raw' disk!

          D, [2011-02-23T18:04:59.488702 #4472] DEBUG -- : /usr/lib/ruby/gems/1.8/gems/boxgrinder-build-0.8.1/lib/boxgrinder-build/helpers/guestfs-helper.rb:242:in `mount_partitions'

          /usr/lib/ruby/gems/1.8/gems/boxgrinder-build-0.8.1/lib/boxgrinder-build/helpers/guestfs-helper.rb:185:in `execute'

          /usr/lib/ruby/gems/1.8/gems/boxgrinder-build-0.8.1/lib/boxgrinder-build/helpers/guestfs-helper.rb:124:in `customize'

          /usr/lib/ruby/gems/1.8/gems/boxgrinder-build-0.8.1/lib/boxgrinder-build/helpers/image-helper.rb:153:in `customize'

          /usr/lib/ruby/gems/1.8/gems/boxgrinder-build-rpm-based-os-plugin-0.0.12/lib/boxgrinder-build-rpm-based-os-plugin/rpm-based-os-plugin.rb:83:in `build_with_appliance_creator'

          /usr/lib/ruby/gems/1.8/gems/boxgrinder-build-rhel-os-plugin-0.0.9/lib/boxgrinder-build-rhel-os-plugin/rhel-plugin.rb:33:in `build_rhel'

          /usr/lib/ruby/gems/1.8/gems/boxgrinder-build-centos-os-plugin-0.0.8/lib/boxgrinder-build-centos-os-plugin/centos-plugin.rb:41:in `execute'

          /usr/lib/ruby/gems/1.8/gems/boxgrinder-build-0.8.1/lib/boxgrinder-build/plugins/base-plugin.rb:130:in `run'

          /usr/lib/ruby/gems/1.8/gems/boxgrinder-build-0.8.1/lib/boxgrinder-build/appliance.rb:108:in `execute_os_plugin'

          /usr/lib/ruby/gems/1.8/gems/boxgrinder-build-0.8.1/lib/boxgrinder-build/appliance.rb:82:in `execute_plugin_chain'

          /usr/lib/ruby/gems/1.8/gems/boxgrinder-build-0.8.1/lib/boxgrinder-build/appliance.rb:93:in `create'

          /usr/lib/ruby/gems/1.8/gems/boxgrinder-build-0.8.1/bin/boxgrinder:63:in `build'

          /usr/lib/ruby/gems/1.8/gems/thor-0.13.6/lib/thor/task.rb:33:in `send'

          /usr/lib/ruby/gems/1.8/gems/thor-0.13.6/lib/thor/task.rb:33:in `run'

          /usr/lib/ruby/gems/1.8/gems/thor-0.13.6/lib/thor/invocation.rb:109:in `invoke'

          /usr/lib/ruby/gems/1.8/gems/thor-0.13.6/lib/thor/invocation.rb:116:in `call'

          /usr/lib/ruby/gems/1.8/gems/thor-0.13.6/lib/thor/invocation.rb:116:in `invoke'

          /usr/lib/ruby/gems/1.8/gems/thor-0.13.6/lib/thor.rb:137:in `start'

          /usr/lib/ruby/gems/1.8/gems/thor-0.13.6/lib/thor/base.rb:378:in `start'

          /usr/lib/ruby/gems/1.8/gems/thor-0.13.6/lib/thor.rb:124:in `start'

          /usr/lib/ruby/gems/1.8/gems/boxgrinder-build-0.8.1/bin/boxgrinder:125

          /usr/bin/boxgrinder:19:in `load'

          /usr/bin/boxgrinder:19

          • 2. Re: CentOS on CentOS image build issues
            goldmann

            Hi Tom,

            1.  Appliance-creator was throwing errors during the RPM install process related to "lsetfilecon."  Putting SELinux on the host into Permissive mode did not work.  I had to disable SELinux entirely on the host.

            We haven't focused on SELinux so far, so yes - this may be an issue. I filled a bug report to track it: https://issues.jboss.org/browse/BGBUILD-170 I cannot guarantee when it'll be fixed

            2.  The version of parted that comes with CentOS is buggy with respect to requesting the kernel to re-read a parition table on a loop device.  I had to upgrade parted to the latest upstream GNU release (2.3) in order to get it to work properly. (I forward ported the patches in the CentOS RPM that had not been upstreamed and remade the parted RPM if anyone wants to take a look at it. There is a problem with .so versioning with the RPM but that can probably be solved easily enough.)

            Hmm, that's new to me. Could you please explain a bit more what the issue is? How this fails the build?

            3.  I believe I've encountered a race condition in the install process where the RPM installation is still proceeding when the code attempts to run the grub install.  I encountered a version of this race condition when I was building a Fedora VM on a Fedora host.  I tracked it to a RPM package with non-trivial %post scripts that was not completing instantly.  I removed that package from the appliance and the boxgrinder failure went away.

             

            Here is relevant snippet of my BoxGrinder log:

             

             

            D, [2011-02-23T18:03:33.234064 #4472] DEBUG -- : Installing: pm-utils                     ##############

            ####### [290/291]

            D, [2011-02-23T18:03:34.298630 #4472] DEBUG -- : Writing grub /home/tdyas/BoxGrinder/build/appliances/x8

            6_64/centos/5/centos-node/centos-plugin/tmp/imgcreate-mDrAaF/install_root/boot/grub/device.map

            D, [2011-02-23T18:03:36.853613 #4472] DEBUG -- : ls: /home/tdyas/BoxGrinder/build/appliances/x86_64/cent

            os/5/centos-node/centos-plugin/tmp/imgcreate-mDrAaF/install_root/boot/initramfs*: No such file o

            r directory

            D, [2011-02-23T18:03:36.856012 #4472] DEBUG -- : Writing grub config /home/tdyas/BoxGrinder/build/applia

            nces/x86_64/centos/5/centos-node/centos-plugin/tmp/imgcreate-mDrAaF/install_root/boot/grub/grub.

            conf

            D, [2011-02-23T18:03:36.856191 #4472] DEBUG -- : Copying /home/tdyas/BoxGrinder/build/appliances/x86_64/

            centos/5/centos-node/centos-plugin/tmp/imgcreate-mDrAaF/install_root/usr/share/grub/x86_64-redha

            t/e2fs_stage1_5 to /home/tdyas/BoxGrinder/build/appliances/x86_64/centos/5/centos-node/centos-pl

            ugin/tmp/imgcreate-mDrAaF/install_root/boot/grub/e2fs_stage1_5

            D, [2011-02-23T18:03:36.868901 #4472] DEBUG -- : Copying /home/tdyas/BoxGrinder/build/appliances/x86_64/

            centos/5/centos-node/centos-plugin/tmp/imgcreate-mDrAaF/install_root/usr/share/grub/x86_64-redha

            t/stage1 to /home/tdyas/BoxGrinder/build/appliances/x86_64/centos/5/centos-node/centos-plugin/tm

            p/imgcreate-mDrAaF/install_root/boot/grub/stage1

            D, [2011-02-23T18:03:36.870879 #4472] DEBUG -- : Copying /home/tdyas/BoxGrinder/build/appliances/x86_64/

            centos/5/centos-node/centos-plugin/tmp/imgcreate-mDrAaF/install_root/usr/share/grub/x86_64-redha

            t/stage2 to /home/tdyas/BoxGrinder/build/appliances/x86_64/centos/5/centos-node/centos-plugin/tm

            p/imgcreate-mDrAaF/install_root/boot/grub/stage2

            D, [2011-02-23T18:03:38.757879 #4472] DEBUG -- : Installing grub to /dev/loop1

            D, [2011-02-23T18:03:38.845241 #4472] DEBUG -- : Probing devices to guess BIOS drives. This may take a l

            ong time.

            D, [2011-02-23T18:03:38.926735 #4472] DEBUG -- : Installing: kudzu                        ##############

            ####### [291/291]

            D, [2011-02-23T18:03:38.926846 #4472] DEBUG -- :

            D, [2011-02-23T18:03:38.926904 #4472] DEBUG -- : GNU GRUB  version 0.97  (640K lower / 3072K upper memor

            y)

            D, [2011-02-23T18:03:38.926954 #4472] DEBUG -- :

            D, [2011-02-23T18:03:38.927005 #4472] DEBUG -- : [ Minimal BASH-like line editing is supported.  For the

            first word, TAB

            D, [2011-02-23T18:03:38.927062 #4472] DEBUG -- : lists possible command completions.  Anywhere else TAB

            lists the possible

            D, [2011-02-23T18:03:38.927116 #4472] DEBUG -- : completions of a device/filename.]

            D, [2011-02-23T18:03:38.927184 #4472] DEBUG -- : grub> device (hd0) /dev/loop1

            D, [2011-02-23T18:03:38.927241 #4472] DEBUG -- : grub> root (hd0,0)

             

            You will note that kudzu installed after the grub stage had already started. Is there any way to make appliance-creator wait for the RPM install stage to complete?

             

            I think that's just a logging problem in BG. Messages are redirected to one output from STDOUT and STERR, which may mess the order of messages sometimes. Created an issue: https://issues.jboss.org/browse/BGBUILD-171

             

            I saw also some post script failing, but s far I cannot tell you why it happens.

             

            Thanks for reports!

             

            --Marek

            • 3. Re: CentOS on CentOS image build issues
              goldmann

              tdyas wrote:

               

              Forgot to include the actual error that killed the build.  Basically, BoxGrinder cannot unmount the loop device and then cannot find the raw disk image.  Here is the remainder of the log file:

               

              D, [2011-02-23T18:03:38.927302 #4472] DEBUG -- : Filesystem type is ext2fs, partition type 0x83

              D, [2011-02-23T18:03:38.927366 #4472] DEBUG -- : grub> setup --stage2=/home/tdyas/BoxGrinder/build/appli

              ances/x86_64/centos/5/centos-node/centos-plugin/tmp/imgcreate-mDrAaF/install_root/boot/grub/stag

              e2 --prefix=/grub  (hd0)

              D, [2011-02-23T18:03:38.927423 #4472] DEBUG -- : Checking if "/grub/stage1" exists... yes

              D, [2011-02-23T18:03:38.927476 #4472] DEBUG -- : Checking if "/grub/stage2" exists... yes

              D, [2011-02-23T18:03:38.927535 #4472] DEBUG -- : Checking if "/grub/e2fs_stage1_5" exists... yes

              D, [2011-02-23T18:03:38.927594 #4472] DEBUG -- : Running "embed /grub/e2fs_stage1_5 (hd0)"... failed (th

              is is not fatal)

              D, [2011-02-23T18:03:38.927741 #4472] DEBUG -- : Running "embed /grub/e2fs_stage1_5 (hd0,0)"... failed (

              this is not fatal)

              D, [2011-02-23T18:03:38.927835 #4472] DEBUG -- : Running "install --stage2=/home/tdyas/BoxGrinder/build/

              appliances/x86_64/centos/5/centos-node/centos-plugin/tmp/imgcreate-mDrAaF/install_root/boot/grub

              /stage2 /grub/stage1 (hd0) /grub/stage2 p /grub/grub.conf "... succeeded

              D, [2011-02-23T18:03:38.927924 #4472] DEBUG -- : Done.

              D, [2011-02-23T18:03:38.928010 #4472] DEBUG -- : grub> quit

              D, [2011-02-23T18:03:43.410515 #4472] DEBUG -- : /sbin/setfiles:  labeling files under /

              D, [2011-02-23T18:03:49.873074 #4472] DEBUG -- : /sbin/setfiles:  labeling files under /

              D, [2011-02-23T18:03:49.873287 #4472] DEBUG -- : matchpathcon_filespec_eval:  hash table stats: 33730 el

              ements, 23879/65536 buckets used, longest chain length 3

              D, [2011-02-23T18:03:49.873579 #4472] DEBUG -- : /sbin/setfiles:  Done.

              D, [2011-02-23T18:03:49.970498 #4472] DEBUG -- : umount: /home/tdyas/BoxGrinder/build/appliances/x86_64/

              centos/5/centos-node/centos-plugin/tmp/imgcreate-mDrAaF/install_root/proc: device is busy

              D, [2011-02-23T18:03:49.972629 #4472] DEBUG -- : umount: /home/tdyas/BoxGrinder/build/appliances/x86_64/

              centos/5/centos-node/centos-plugin/tmp/imgcreate-mDrAaF/install_root/proc: device is busy

              D, [2011-02-23T18:03:49.975505 #4472] DEBUG -- : Unmounting directory /home/tdyas/BoxGrinder/build/appli

              ances/x86_64/centos/5/centos-node/centos-plugin/tmp/imgcreate-mDrAaF/install_root/boot

              D, [2011-02-23T18:03:49.981355 #4472] DEBUG -- : Unmounting directory /home/tdyas/BoxGrinder/build/appli

              ances/x86_64/centos/5/centos-node/centos-plugin/tmp/imgcreate-mDrAaF/install_root/

              Was it a clean build? Did the build fail in earlier stage earlier? Is it reproducible, also fter restart? If yes, could you please send me your .appl file and host info (Ruby version, appliance-creator version)?

               

              I see this message for the first time.

               

              --Marek

              • 4. CentOS on CentOS image build issues
                tdyas

                I solved the problem.  I narrowed the problematic RPM package down to the Sun Java "jdk" package.  Searching this forum for "jdk," I saw that a work-around was posted last year (http://community.jboss.org/message/560844).  Could this go in the FAQ?

                 

                Regarding parted, I tested again using BG with the CentOS version of parted (but without jdk). BG was able to build an image, although it did print out these error messages:

                 

                D, [2011-02-25T15:54:23.203334 #9566] DEBUG -- : Formatting disks

                D, [2011-02-25T15:54:23.204007 #9566] DEBUG -- : Initializing partition table for /dev/loop2

                D, [2011-02-25T15:54:23.316217 #9566] DEBUG -- : Assigning partitions to disks

                D, [2011-02-25T15:54:23.316442 #9566] DEBUG -- : Assigned /boot to sda1 at 0 at size 102

                D, [2011-02-25T15:54:23.316871 #9566] DEBUG -- : Assigned / to sda2 at 102 at size 4096

                D, [2011-02-25T15:54:23.317043 #9566] DEBUG -- : Creating partitions

                D, [2011-02-25T15:54:23.317260 #9566] DEBUG -- : Add primary part at 0 of size 102

                D, [2011-02-25T15:54:23.344368 #9566] DEBUG -- : Error: Error informing the kernel about modifications to partition /dev/loop2p1 -- Invalid argument. This means Linux won't know about any changes you made to /dev/loop2p1 until you reboot -- so you shouldn't mount it or use it in any way before rebooting.

                D, [2011-02-25T15:54:23.344474 #9566] DEBUG -- : Warning: The kernel was unable to re-read the partition table on /dev/loop2 (Invalid argument). This means Linux won't know anything about the modifications you made until you reboot.  You should reboot your computer before doing anything with /dev/loop2.

                D, [2011-02-25T15:54:23.344772 #9566] DEBUG -- : Add primary part at 102 of size 4096

                D, [2011-02-25T15:54:23.560985 #9566] DEBUG -- : Error: Error informing the kernel about modifications to partition /dev/loop2p1 -- Invalid argument. This means Linux won't know about any changes you made to /dev/loop2p1 until you reboot -- so you shouldn't mount it or use it in any way before rebooting.

                D, [2011-02-25T15:54:23.561155 #9566] DEBUG -- : Error: Error informing the kernel about modifications to partition /dev/loop2p2 -- Invalid argument.  This means Linux won't know about any changes you made to /dev/loop2p2 until you reboot -- so you shouldn't mount it or use it in any way before rebooting.

                D, [2011-02-25T15:54:23.561267 #9566] DEBUG -- : Warning: The kernel was unable to re-read the partition table on /dev/loop2 (Invalid argument). This means Linux won't know anything about the modifications you made until you reboot.  You should reboot your computer before doing anything with /dev/loop2.

                D, [2011-02-25T15:54:23.566977 #9566] DEBUG -- : Dev loop2p1: /dev/loop21 -> /dev/mapper/loop2p1

                D, [2011-02-25T15:54:23.567098 #9566] DEBUG -- : Dev loop2p2: /dev/loop22 -> /dev/mapper/loop2p2

                D, [2011-02-25T15:54:23.567225 #9566] DEBUG -- : Adding partx mapping for /dev/loop2

                 

                I upgraded to parted 2.3 from the GNU website and the error messages went away.

                • 5. CentOS on CentOS image build issues
                  goldmann

                  Hi Tom,

                  tdyas wrote:

                   

                  I solved the problem.  I narrowed the problematic RPM package down to the Sun Java "jdk" package.  Searching this forum for "jdk," I saw that a work-around was posted last year (http://community.jboss.org/message/560844).  Could this go in the FAQ?

                  Yes, I'll add it to FAQ shortly. Thank you for spotting this out!

                   

                  Regarding parted, I tested again using BG with the CentOS version of parted (but without jdk). BG was able to build an image, although it did print out these error messages:

                   

                  D, [2011-02-25T15:54:23.203334 #9566] DEBUG -- : Formatting disks

                  D, [2011-02-25T15:54:23.204007 #9566] DEBUG -- : Initializing partition table for /dev/loop2

                  D, [2011-02-25T15:54:23.316217 #9566] DEBUG -- : Assigning partitions to disks

                  D, [2011-02-25T15:54:23.316442 #9566] DEBUG -- : Assigned /boot to sda1 at 0 at size 102

                  D, [2011-02-25T15:54:23.316871 #9566] DEBUG -- : Assigned / to sda2 at 102 at size 4096

                  D, [2011-02-25T15:54:23.317043 #9566] DEBUG -- : Creating partitions

                  D, [2011-02-25T15:54:23.317260 #9566] DEBUG -- : Add primary part at 0 of size 102

                  D, [2011-02-25T15:54:23.344368 #9566] DEBUG -- : Error: Error informing the kernel about modifications to partition /dev/loop2p1 -- Invalid argument. This means Linux won't know about any changes you made to /dev/loop2p1 until you reboot -- so you shouldn't mount it or use it in any way before rebooting.

                  D, [2011-02-25T15:54:23.344474 #9566] DEBUG -- : Warning: The kernel was unable to re-read the partition table on /dev/loop2 (Invalid argument). This means Linux won't know anything about the modifications you made until you reboot.  You should reboot your computer before doing anything with /dev/loop2.

                  D, [2011-02-25T15:54:23.344772 #9566] DEBUG -- : Add primary part at 102 of size 4096

                  D, [2011-02-25T15:54:23.560985 #9566] DEBUG -- : Error: Error informing the kernel about modifications to partition /dev/loop2p1 -- Invalid argument. This means Linux won't know about any changes you made to /dev/loop2p1 until you reboot -- so you shouldn't mount it or use it in any way before rebooting.

                  D, [2011-02-25T15:54:23.561155 #9566] DEBUG -- : Error: Error informing the kernel about modifications to partition /dev/loop2p2 -- Invalid argument.  This means Linux won't know about any changes you made to /dev/loop2p2 until you reboot -- so you shouldn't mount it or use it in any way before rebooting.

                  D, [2011-02-25T15:54:23.561267 #9566] DEBUG -- : Warning: The kernel was unable to re-read the partition table on /dev/loop2 (Invalid argument). This means Linux won't know anything about the modifications you made until you reboot.  You should reboot your computer before doing anything with /dev/loop2.

                  D, [2011-02-25T15:54:23.566977 #9566] DEBUG -- : Dev loop2p1: /dev/loop21 -> /dev/mapper/loop2p1

                  D, [2011-02-25T15:54:23.567098 #9566] DEBUG -- : Dev loop2p2: /dev/loop22 -> /dev/mapper/loop2p2

                  D, [2011-02-25T15:54:23.567225 #9566] DEBUG -- : Adding partx mapping for /dev/loop2

                   

                  I upgraded to parted 2.3 from the GNU website and the error messages went away.

                  I get used to this warnings The warnings doesn't mean that something went wrong. You can ignore them. Thanks for the info on parted 2.3, but I cannot solve it - I need to wait for update in CentOS repos. Maybe CentOS 6 will fix it? It'll contain parted-2.1.

                   

                  --Marek