Summary: Mounting device files prevents using systemd-nspawn in mock Key: OVIRT-2032 URL: https://ovirt-jira.atlassian.net/browse/OVIRT-2032 Project: oVirt - virtualization made easy Issue Type: Bug Components: mock_runner Reporter: Barak Korren Assignee: infra
The systemd-nspawn functionality in mock is introduced in OVIRT-2031.
When using systemd-nspawn mock uses some kind of a layered FS that is created when it starts and removed when it exits. This is different from the way it works when using chroot where the directory that the chroot is configured in stays around until it is explicitly removed.
mock had an issue where if you tried to bind-mount something into the mock environment, you had to have the mount point ready for it. We worked around this in {{mock_runner}} by using the fact the chroot was persistent, going into it and setting up the mount points as needed before actually starting to use it to run the STDCI script.
Since the mock authors were aware of the issue when the implemented the systemd-nspawn functionality, they made mock pre-create mount point in the container as needed. However, they only supported directory mount points, so if we tried to bind-mount a socket we would get en error message failing to mount a socket file on a directory.
We reported this issue to the mock developers as issue [#87|https://github.com/rpm-software-management/mock/issues/87]
— This message was sent by Atlassian Jira (v1001.0.0-SNAPSHOT#100085)