This was an interesting failure for folks working with docker.

Build 119 was launched manually.
Build 120 was scheduled to run automatically, but was blocked until build 119 completed.
Build 119 completed, and triggered an asynchronous cleanup of the docker container, and image used for building.
Build 120 is not longer blocked, and stars setting up a docker image for it's build.
The docker daemon sees reusable layers from the 119 image and tries to use them for the 120 image, but is seemingly unaware that the 119 image is being deleted along with all it's layers (it seem to think their are no other users of those layers.

So there is some sort of race/lock issue in the docker daemon when it comes to layer management.

$docker --version
Docker version 18.09.5, build e8ff056

For now I'll use a jenkins lockable resource to ensure image setup/teardown are atomic.

Scott



On 2019-05-28 9:52 p.m., build.starlingx@gmail.com wrote:
Project: STX_build_master_master
Build #: 120
Status: Failure
Timestamp: 20190529T014905Z

Check logs at:
http://mirror.starlingx.cengn.ca/mirror/starlingx/master/centos/20190529T014905Z/logs
--------------------------------------------------------------------------------
Parameters

BUILD_CONTAINERS_DEV: false
BUILD_CONTAINERS_STABLE: false

_______________________________________________
Starlingx-discuss mailing list
Starlingx-discuss@lists.starlingx.io
http://lists.starlingx.io/cgi-bin/mailman/listinfo/starlingx-discuss