Drop separate cpupinning jobs

For all the tests that actually run upstream, we've completed the
conversion to pre-deployed CPU pinning configuration, and are no
longer changing cpu_(dedicated|shared)_set onthe fly. We can re-merge
all the jobs into a single main job. Let's also use this opportunity
to drop the -legacy job that tested with vcpu_pin)_set, as that's no
longer relevant upstream.

Change-Id: Ifac17163876b770a2ff3cc6c477256f1e8732553
This commit is contained in:
Artom Lifshitz 2021-12-03 13:50:16 -05:00
parent f5b8464534
commit 485d1dd2bc

View File

@ -29,8 +29,7 @@
- compute
- job:
name: whitebox-devstack-multinode-base
abstract: true
name: whitebox-devstack-multinode
parent: tempest-multinode-full-py3
nodeset: nested-virt-multinode
description: |
@ -57,6 +56,9 @@
devstack_local_conf:
post-config:
$NOVA_CONF:
compute:
cpu_dedicated_set: '0-3'
cpu_shared_set: '4,5'
libvirt:
cpu_mode: custom
cpu_models: Nehalem
@ -71,6 +73,9 @@
devstack_local_conf:
post-config:
$NOVA_CONF:
compute:
cpu_dedicated_set: '4-7'
cpu_shared_set: '2,3'
libvirt:
cpu_mode: custom
cpu_models: Nehalem
@ -80,65 +85,9 @@
tempest:
num_hugepages: 512
- job:
name: whitebox-devstack-multinode
parent: whitebox-devstack-multinode-base
vars:
# NOTE(artom) We can't have this on the parent job, otherwise the two
# -cpupinnig jobs will inherit it as well.
tempest_exclude_regex: 'test_live_migrate_and_reboot|test_shared_pinned_and_unpinned_guest|^whitebox_tempest_plugin.api.compute.test_cpu_pinning.EmulatorThreadTest|test_cpu_pinning_and_emulator_threads|test_collocation_migration'
- job:
name: whitebox-devstack-multinode-cpupinning
parent: whitebox-devstack-multinode-base
description: |
Runs the CPU pinning tests on single-NUMA, non-SMT, nested virt VMs. Uses
[compute]cpu_dedicated_set to configure host CPUs for pinning.
vars:
tempest_test_regex: 'test_live_migrate_and_reboot|test_shared_pinned_and_unpinned_guest|^whitebox_tempest_plugin.api.compute.test_cpu_pinning.EmulatorThreadTest|test_cpu_pinning_and_emulator_threads|test_collocation_migration'
devstack_local_conf:
post-config:
$NOVA_CONF:
compute:
cpu_dedicated_set: '0-3'
cpu_shared_set: '4,5'
group-vars:
subnode:
devstack_local_conf:
post-config:
$NOVA_CONF:
compute:
cpu_dedicated_set: '4-7'
cpu_shared_set: '2,3'
- job:
name: whitebox-devstack-multinode-cpupinninglegacy
parent: whitebox-devstack-multinode-base
description: |
Runs the CPU pinning tests on single-NUMA, non-SMT, nested virt VMs. Uses
[DEFAULT]vcpu_pin_set to configure host CPUs for pinning.
vars:
tempest_test_regex: 'test_live_migrate_and_reboot|^whitebox_tempest_plugin.api.compute.test_cpu_pinning.EmulatorThreadTest'
devstack_local_conf:
post-config:
$NOVA_CONF:
DEFAULT:
vcpu_pin_set: '0-3'
compute:
cpu_shared_set: '4,5'
group-vars:
subnode:
devstack_local_conf:
post-config:
$NOVA_CONF:
DEFAULT:
vcpu_pin_set: '4-7'
compute:
cpu_shared_set: '2,3'
- job:
name: whitebox-devstack-ceph-compute-local-ephemeral
parent: whitebox-devstack-multinode-base
parent: whitebox-devstack-multinode
nodeset: nested-virt-multinode
description: |
Runs test_rbd_direct_download test on single-NUMA, non-SMT, nested virt
@ -148,7 +97,7 @@
storage. It also enables the direct download of images via rbd into
the local imagecache for Nova.
required-projects:
- openstack/devstack-plugin-ceph
- openstack/devstack-plugin-ceph
irrelevant-files:
- ^whitebox_tempest_plugin/api/compute/(?!test_rbd_direct_download).*py$
vars:
@ -190,13 +139,8 @@
check:
jobs:
- whitebox-devstack-multinode
- whitebox-devstack-multinode-cpupinning
- whitebox-devstack-multinode-cpupinninglegacy
# TODO(amodi): Add a periodic job for below
- whitebox-devstack-ceph-compute-local-ephemeral
gate:
jobs:
- whitebox-devstack-multinode
- whitebox-devstack-multinode-cpupinning
- whitebox-devstack-multinode-cpupinninglegacy
- whitebox-devstack-ceph-compute-local-ephemeral