ff9c2f782f
After a forced reboot, kubelet fails to clean up orphaned volume files and directories. These orphaned volumes persist because the directories still hold files, a situation that would typically be resolved during a normal flow but instead is caused by the forced reboot. The volume directories are tied to the pods' unique identifiers. Since none of the pods are running after the reboot, the associated volume directories and files can be safely removed. The work in this commit will eventually be superseded by https://github.com/kubernetes/enhancements/issues/3756. Test Plan: PASS: - Build an iso and install an aio-sx system and verify that the cleanup script is installed and the kubelet service is running with ExecStarPre=/usr/local/bin/kubelet-cleanup-orphaned-volumes.sh. PASS: - Reboot the system with active pods that contain files in their volumes directories and verify that all volume directories and their files under /var/lib/kubelet/pods/ are deleted after reboot. PASS: - Verify that explictly restarting the kubelet service does not attempt to delete kubelet volume directories. PASS: - Verify volume-subpaths directories and files are cleaned up after reboot. Closes-Bug: 2027810 Change-Id: Ie7e637c4d5e79ec08d33bd80dade35890b711548 Signed-off-by: Gleb Aronsky <gleb.aronsky@windriver.com>