Fix legacy-grenade-dsvm-neutron-multinode-live-migration

Since change I81301eeecc7669a169deeb1e2c5d298a595aab94 in
devstack, nova-cpu.conf is a copy of nova.conf minus
database access. Grenade jobs also run nova-compute with
nova-cpu.conf anyway so we can just drop the conditional
which was otherwise messing up the config file that the
ceph script would write rbd configuration which is why
live block migration tests with ceph were failing.

NOTE(mriedem): This is not a pure backport since this change
in queens just disables running the job with ceph because of
a mess of changes to devstack queens related to change
https://review.openstack.org/625131/ and change
https://review.openstack.org/632100/. As a result, the pike
node runs with nova.conf (because of grenade singleconductor
mode makes it so) but the queens node runs with nova-cpu.conf
and then the _ceph_configure_nova function does not configure
the nodes for rbd auth properly which makes rbd-backed live
migration in the job fail. Rather than try to sort that all out
with a one-off change in queens (which is pretty old at this
point anyway), this change just skips the ceph portion of the
job. An alternative is just not running the job in queens, but
we can easily still get block migrate live migration coverage
so we might as well keep that running.

Change-Id: I902e459093af9b82f9033d58cffcb2a628f5ec39
Closes-Bug: #1813216
(cherry picked from commit f2db43d57a)
(cherry picked from commit c4a2a32f07)
This commit is contained in:
Matt Riedemann 2019-02-05 10:40:33 -05:00
parent 12c7aa3994
commit 5d5e7be8cb
1 changed files with 17 additions and 15 deletions

View File

@ -41,23 +41,25 @@ echo '2. NFS testing is skipped due to setup failures with Ubuntu 16.04'
#run_tempest "NFS shared storage test" "live_migration"
#nfs_teardown
echo '3. test with Ceph for root + ephemeral disks'
prepare_ceph
GLANCE_API_CONF=${GLANCE_API_CONF:-/etc/glance/glance-api.conf}
configure_and_start_glance
# Deal with grenade craziness...
if [ "$GRENADE_OLD_BRANCH" ]; then
# NOTE(mriedem): Grenade runs in singleconductor mode, so it won't use
# /etc/nova/nova-cpu.conf so we have to overwrite NOVA_CPU_CONF which is
# read in configure_and_start_nova.
if ! is_service_enabled n-super-cond; then
NOVA_CPU_CONF=$NOVA_CONF
fi
fi
# NOTE(mriedem): Testing with ceph in the grenade live migration job is
# disabled because of a mess of changes in devstack from queens which
# result in the pike node running with nova.conf and the queens node
# running with nova-cpu.conf and _ceph_configure_nova (in ceph.sh) does
# not configure the nodes properly for rbd auth which makes rbd-backed
# live migration fail (because the nodes on shared storage can't
# communicate). Fixing that is non-trivial so we just skip ceph testing
# in the grenade case.
echo '2. test with Ceph is skipped due to bug 1813216'
else
echo '3. test with Ceph for root + ephemeral disks'
prepare_ceph
GLANCE_API_CONF=${GLANCE_API_CONF:-/etc/glance/glance-api.conf}
configure_and_start_glance
configure_and_start_nova
run_tempest "Ceph nova&glance test" "^.*test_live_migration(?!.*(test_volume_backed_live_migration))"
configure_and_start_nova
run_tempest "Ceph nova&glance test" "^.*test_live_migration(?!.*(test_volume_backed_live_migration))"
fi
set +e
#echo '4. test with Ceph for volumes and root + ephemeral disk'