baremetal, virtual: Bump kernel to hwe-18.04 (5.0) 25/69525/4
authorAlexandru Avadanii <Alexandru.Avadanii@enea.com>
Wed, 8 Jan 2020 16:33:13 +0000 (17:33 +0100)
committerAlexandru Avadanii <Alexandru.Avadanii@enea.com>
Thu, 9 Jan 2020 09:10:18 +0000 (10:10 +0100)
commitf03bfd30206c1d21de7e5c9ef2f6ed79f6ee13f4
treec3e28dd2cd07e2d681d9a075c313b310dfaa13fe
parent5527711008fdd6cb2558f18d901c40efbc5d522b
baremetal, virtual: Bump kernel to hwe-18.04 (5.0)

On some aarch64 platforms (e.g. ThunderX 1), lvcreate manifests some
spurious timing issues resulting in incomplete/corrupted LVM thin

creation and eventually to transaction ID mismatch between userspace
and kernel space.

This eventually leads to cinder-volume issues, either when creating
the thin storage pool (vgroot-pool) and/or when creating the LVs
inside said pool.

The issue manifests spuriously on Ubuntu Bionic + UCA, so until a
working combination of userspace/kernel is found, work around this
by bumping the kernel package to hwe-18.04 (kernel 5.0),
effectively bypassing the timing issues during volume creation.

This affects all cluster machines (both HA and NOHA scenarios,
baremetal and virtual, x86_64 and aarch64, baremetal and virtualized
nodes).

Note: Ubuntu Bionic cloud image partition handling requires e2fsprogs
1.43, not currently available on Ubuntu Xenial / CentOS 7.

Change-Id: I839e03080104c391fe18185b9544c9df43c114e6
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
ci/deploy.sh
mcp/config/scenario/defaults.yaml.j2
mcp/reclass/classes/cluster/all-mcp-arch-common/infra/maas.yml.j2
mcp/scripts/lib_jump_common.sh
mcp/scripts/lib_jump_deploy.sh
mcp/scripts/requirements_deb.yaml