http://bugzilla.opensuse.org/show_bug.cgi?id=982097 Bug ID: 982097 Summary: docker: non-standard package update warning Classification: openSUSE Product: openSUSE Distribution Version: Leap 42.1 Hardware: Other OS: Other Status: NEW Severity: Major Priority: P5 - None Component: Maintenance Assignee: fcastelli@suse.com Reporter: astieger@suse.com QA Contact: qa-bugs@suse.de CC: bbrunner@suse.com, fcastelli@suse.com, maintenance@opensuse.org Found By: --- Blocker: --- The proposed docker update in openSUSE:Maintenance:5092 shows the following message: $ rpm -q docker docker-1.9.1-13.1.x86_64 $ zypper up docker Loading repository data... Reading installed packages... Resolving package dependencies... The following NEW package is going to be installed: docker-zsh-completion The following package is going to be upgraded: docker 1 package to upgrade, 1 new. Overall download size: 0 B. Already cached: 7.0 MiB. After the operation, additional 10.7 MiB will be used. Continue? [y/n/? shows all options] (y): y In cache docker-1.11.1-18.1.x86_64.rpm (1/2), 6.8 MiB ( 33.5 MiB unpacked) In cache docker-zsh-completion-1.11.1-18.1.noarch.rpm (2/2), 120.4 KiB ( 50.8 KiB unpacked) Checking for file conflicts: ...........................................................................[done] (1/2) Installing: docker-1.11.1-18.1.x86_64 ...........................................................[error] Installation of docker-1.11.1-18.1.x86_64 failed: Error: Subprocess failed. Error: RPM failed: *** WARNING *** In the migration from docker<1.10.0 to docker>=1.10.0, the Docker image format has changed to be completely content-addressible. This results in several positive improvements to image operations (better caching during builds mainly). However, the migration operation may take several hours if you have a lot of large images on a Docker host. In order to ensure that you have minimum downtime, this update of Docker will not complete successfully, and you will have the opportunity to run a separate migration tool (which will not cause downtime for your Docker daemon). In order to run this migration tool, please install the 'docker-image-migrator' package. You can run the migration with this command, which will exit after the migration has been completed: $ /usr/lib/docker-image-migrator/do-image-migration-v1to2.sh Because the migrator requires information about the storage driver used by Docker, the migration script will source /etc/sysconfig/docker and use $DOCKER_OPTS as arguments to the migrator. If this automated migration fails, it will be re-attempted with every known storage driver. In addition, the script accepts arguments which will simiarly be appended to the set of arguments (after $DOCKER_OPTS) to the migrator. However, if you prefer to not run this separate migration tool, you can force this update using the following command. THIS WILL CAUSE DOWNTIME, BECAUSE DOCKER WILL RUN THE MIGRATION ON FIRST START AND YOU WILL BE UNABLE TO START ANY CONTAINERS OR USE ANY DOCKER COMMANDS (EVEN CONTAINERS WITH RESTART POLICIES ACTIVE): $ DOCKER_FORCE_INSTALL=1 sudo -E zypper up docker error: %pre(docker-1.11.1-18.1.x86_64) scriptlet failed, exit status 1 error: docker-1.11.1-18.1.x86_64: install failed error: docker-1.9.1-13.1.x86_64: erase skipped ------------- Failing an rpm subprocess like this breaks all kinds of third party tools and GUI tools. You are introducing a single-purpose environment variable? Would such downtime considerations not be better suited for the patchinfo? I do not think that it should fail like this. Can you not bring the package into a state where such migration is delayed? If this is not possible and maintenance update is expected to break things, this should not be a maintenance update (new docker version for the next Leap release) -- You are receiving this mail because: You are on the CC list for the bug.