CI: periodic jobs fail on undercloud install because "AttributeError: 'module' object has no attribute 'RegionInvalidationStrategy'"

[Sagi (Sergey) Shnaidman]

undercloud installation fails because fail of keystone-manage db_sync command:

  File "/usr/bin/keystone-manage", line 6, in <module>
    from keystone.cmd.manage import main
  File "/usr/lib/python2.7/site-packages/keystone/cmd/manage.py", line 31, in <module>
    from keystone.cmd import cli
  File "/usr/lib/python2.7/site-packages/keystone/cmd/cli.py", line 30, in <module>
    from keystone.common import openssl
  File "/usr/lib/python2.7/site-packages/keystone/common/openssl.py", line 21, in <module>
    from keystone.common import utils
  File "/usr/lib/python2.7/site-packages/keystone/common/utils.py", line 37, in <module>
    from keystone.common import authorization
  File "/usr/lib/python2.7/site-packages/keystone/common/authorization.py", line 23, in <module>
    from keystone.models import token_model
  File "/usr/lib/python2.7/site-packages/keystone/models/token_model.py", line 22, in <module>
    from keystone.federation import constants
  File "/usr/lib/python2.7/site-packages/keystone/federation/__init__.py", line 15, in <module>
    from keystone.federation.core import * # noqa
  File "/usr/lib/python2.7/site-packages/keystone/federation/core.py", line 17, in <module>
    from keystone.common import cache
  File "/usr/lib/python2.7/site-packages/keystone/common/cache/__init__.py", line 15, in <module>
    from keystone.common.cache.core import * # noqa
  File "/usr/lib/python2.7/site-packages/keystone/common/cache/core.py", line 56, in <module>
    class DistributedInvalidationStrategy(region.RegionInvalidationStrategy):
AttributeError: 'module' object has no attribute 'RegionInvalidationStrategy'
2016-08-30 06:29:35.000 | 2016-08-30 06:29:35 - Error: keystone-manage db_sync returned 1 instead of one of [0]
2016-08-30 06:29:35.000 | 2016-08-30 06:29:35 - Error: /Stage[main]/Keystone::Db::Sync/Exec[keystone-manage db_sync]/returns: change from notrun to 0 failed: keystone-manage db_sync returned 1 instead of one of [0]

http://logs.openstack.org/periodic/periodic-tripleo-ci-centos-7-ovb-ha/c4886b3/logs/undercloud/var/log/undercloud_install.txt.gz#_2016-08-30_06_25_37_000

http://logs.openstack.org/periodic/periodic-tripleo-ci-centos-7-ovb-nonha/37fe6d7/logs/undercloud/var/log/undercloud_install.txt.gz#_2016-08-30_06_29_34_000

Affects: all periodic jobs

[Emilien Macchi (emilienm)]

I don't have all context about the actual problem, but all I know is that it's fixed in python-dogpile-cache-0.6.2-1.el7.noarch.rpm that is now in buildlogs repository so CI should be green again and promotion ok.

[Javier Peña (jpena-c)]

The issue is caused by https://github.com/openstack/keystone/commit/42eda48c78f1153081b4c193dc13c88561409fd3 . In keystone/common/cache/core.py, it does the following:

from dogpile.cache import region

...

class DistributedInvalidationStrategy(region.RegionInvalidationStrategy):

dogpile.cache.region.RegionInvalidationStrategy was added to dogpile.cache in version 0.6.2 (https://bitbucket.org/zzzeek/dogpile.cache/commits/d521db7eb4eab4ad2bacf5f1f5fd89bd3e43a0fa). Previously, we only had version 0.6.1 in our repos, based on the keystone requirements, so it is in fact a Keystone bug because they were using a functionality present in a later release without bumping their requirements.

The requirements issue has been fixed already via https://review.openstack.org/362795. We now have python-dogpile-cache-0.6.2-1.el7.noarch.rpm in the buildlogs repo, so we everything should be fixed.