DMS (Distribution Migration System) Migration failed

SUSE Linux Enterprise Server 15
SUSE Linux Enterprise Server 12

The Distribution Migration System provides an upgrade path for an installed SUSE Linux Enterprise system or SUSE Linux Enterprise Server for SAP Applications from one major version to another. For example, from SUSE Linux Enterprise Server 12 SP5 to SUSE Linux Enterprise Server 15 SP1.

zypper multiple kernel feature

One possible cause of migration failure is due to the zypper multiple kernel feature being enabled, which leaves multiple kernels on the system. Normally this is correct and acceptable, however, due to a kernel RPM script change in certain kernel packages, this is causing a conflict and the older kernels are unable to be removed. 

The DMS procedure may fail completely or partially, depending on where in the process the old/previous kernel is removed.

Look in the log "/var/log/distro_migration.log" for "Migration failed".

These types of errors are seen when the kernel is being removed:

( 928/1204) Removing kernel-default-4.12.14-122.106.1.x86_64 [....error]
Removal of (24555)kernel-default-4.12.14-122.106.1.x86_64(@System) failed:
Error: Subprocess failed. Error: RPM failed: /var/tmp/rpm-tmp.VECCEi: line 1: /usr/lib/module-init-tools/kernel-scriptlets/rpm-preun: No such file or directory
error: %preun(kernel-default-4.12.14-122.106.1.x86_64) scriptlet failed, exit status 127
error: kernel-default-4.12.14-122.106.1.x86_64: erase failed


Abort, retry, ignore? [a/r/i] (a): a
Warning: %posttrans scripts skipped while aborting:
    coreutils-8.29-2.12.x86_64.rpm
    kernel-firmware-20200107-3.15.1.noarch.rpm
    systemd-presets-common-SUSE-15-6.10.noarch.rpm
    kmod-25-6.7.1.x86_64.rpm
    blog-2.18-4.11.x86_64.rpm
    systemd-presets-branding-SLE-15.1-20.5.1.noarch.rpm
    ca-certificates-mozilla-2.44-4.29.1.noarch.rpm
    kbd-2.0.4-8.3.1.x86_64.rpm
    dbus-1-1.12.2-8.3.1.x86_64.rpm
    udev-234-24.64.1.x86_64.rpm
    rpm-4.14.1-10.19.8.x86_64.rpm
    google-opensans-fonts-1.0-1.21.noarch.rpm
    dejavu-fonts-2.37-1.21.noarch.rpm
    dracut-044.2-18.70.1.x86_64.rpm
    python2-chardet-3.0.4-3.23.noarch.rpm
    yp-tools-4.2.2-3.16.x86_64.rpm
    libopenblas_pthreads0-0.2.20-11.11.x86_64.rpm
    sg3_utils-1.44~763+19.1ed0757-9.3.1.x86_64.rpm
    haveged-1.9.2-6.1.x86_64.rpm
    thin-provisioning-tools-0.7.5-1.24.x86_64.rpm
    python2-setuptools-40.5.0-6.3.1.noarch.rpm
    crash-kmp-default-7.2.1_k4.12.14_197.56-9.7.1.x86_64.rpm

Problem occurred during or after installation or removal of packages:
Installation has been aborted as directed.
Please see the above error message for a hint.

Migration failed.

The migration to the new service pack has failed. The system is most
likely in an inconsistent state.

This may leave the system not migrated at all, or only partially migrated.

If most updates are installed, then a simple "zypper update" after a reboot should finalize the migration.

Incorrect product defined in /etc/sle-migration-service.yml

If for example the base product is SLES_SAP but one defined SLES in /etc/sle-migration-service.yml, for example this way:

migration_product: 'SLES/15.1/x86_64'

the migration will fail and one can see similar output in /var/log/distro_migration.log:

Calling: ['chroot', '/system-root', 'SUSEConnect', '--list-extensions']
command 'zypper --disable-repositories --xmlout --non-interactive products -i' failed
Error: zypper returned (7) with '<?xml version='1.0'?>
<stream>
<message type="error">System management is locked by the application with pid 1781 (zypper).

SUSEConnect timeout

SUSEConnect failed to connect to scc.suse.com.

Calling: ['mount', '--bind', '/system-root/usr/lib/zypp/plugins/services', '/usr/lib/zypp/plugins/services']
Calling: ['chroot', '/system-root', 'SUSEConnect', '--list-extensions']
SUSEConnect error: Errno::ETIMEDOUT: Connection timed out - connect(2) for "scc.suse.com" port 443

 

When changing the server registration to use SCC, the server keeps partially reverting back to susecloud.net after a reboot

DMS fails

zypper multiple kernel feature

Engineering is aware of this situation and is investigating a resolution.

Until then the workaround is to manually remove extra RPMs before starting the DMS migration.

If the system is still showing as SLES12-SP5, then all kernel RPMs except the current running kernel will need to be removed prior to running the DMS procedure again.

To do that, check for current installed kernels:

rpm -qa|grep kernel-default

Check the current running kernel:

uname -r

Then remove any kernels that are older.

Here is an example:

rpm -qa|grep kernel-default
kernel-default-4.12.14-122.110.1.x86_64
kernel-default-4.12.14-122.83.1.x86_64
 
uname -r
4.12.14-122.110-default

The old kernel version is "kernel-default-4.12.14-122.83.1.x86_64".

Remove the old kernel:

zypper rm kernel-default-4.12.14-122.83.1

If in doubt, contact SUSE Support for further assistance.
https://www.suse.com/support/kb/doc/?id=000019452
 

Incorrect product defined in /etc/sle-migration-service.yml

Correcting the migration product should make the migration work as expected.  The migration_product must be 'SLES/15.1/x86_64' for SLES migrations and 'SLES_SAP/15.1/x86_64' for SLES_SAP. Changing the product target from the 15-SP1 release is not officially supported.

 

SUSEConnect timeout

The solution could vary depending on cause, but a known issue related to customer need to use HTTP proxy to connect to the internet.

Engineering is aware of incomplete handling of /etc/sysconfig/proxy inside DMS environment. The workaround is to create a shell wrapper for /usr/sbin/SUSEConnect which would explicitly define http_proxy environment variables later passed to original SUSEConnect script.

When changing the server registration to use SCC, the server keeps partially reverting back to susecloud.net after a reboot

Remove the cloudregister cache files and delete any 'susecloud.net' entries from /etc/hosts.  Commenting the entry is not sufficient.  The entries must be deleted.

# Remove cloudregister cache files
rm -rf /var/cache/cloudregister/* 

zypper multiple kernel feature

Due to a kernel RPM script change in certain kernel packages, this is causing a conflict and the older kernels are unable to be removed.

 

Incorrect product defined in /etc/sle-migration-service.yml

It is a misconfiguration since by default no 'migration_product' is set, and the migration will use that product as found in '/etc/products.d/baseproduct'.
 

SUSEConnect timeout

As for proxy related issue, since SUSEConnect is called from a script started via a systemd unit, no shell proxy related environment values are available by default because the systemd unit does not use 'EnvironmentFile=-/etc/sysconfig/proxy' setting.

 

When changing the server registration to use SCC, the server keeps partially reverting back to susecloud.net after a reboot

The cloudregister cache is still populated

• Using the SUSE Distribution Migration System
• Temporarily Disabling Kernel Multiversion Support
• XFS "barrier/nobarrier" mount options have been completely deprecated starting from SLES15 SP2