The tgtd process was not being started in the cinder-volume container.
This results in iSCSI not being avialable for Cinder to operate
against. Unfortunately cinder doesn't report the problem because LVM
is responsible for the iSCSI mapping.
Co-Authored-By: Ryan Hallisey <rhallise@redhat.com>
Closes-Bug: 1473090
Closes-Bug: 1469323
Change-Id: Ie3403c5c3f81beb48960e838b4274e0c5796e387
Many of these were actually files containing the path to the real
script, instead of being symlinks.
config-{internal,external}.sh files in openvswitch/ovs-base were not
needed and were thus deleted.
Also update minimal-environment-vars.md after symlinks were fixed.
Change-Id: I11c3e735067b1feb83c3184c32333e598fbdf590
Closes-Bug: #1474227
Fixed docker client to use ENV if exists to support boot2docker.
Fixed booleans not working as thought with argParser.
Change-Id: I232ed78443199ce20f4b38e12c861c0f97d55c99
Partially-Implements: blueprint build-script
The commands used to create a temporary file on the localhost were
vulnerable to a symlink attack. Removing the shell module and ensuring
the ansible copy and file module is used will verify this file exists as a
file with the correct permissions and ownership
Change-Id: I829483edf1435e41726ebfe1bc826e0c2e5265e3
Closes-Bug: 1471376
Since the volume driver is only accessed through one process, we
probably don't need concurrency locking.
Change-Id: I50f0cb4a7ef92a5fefe5fafb9f3ddadf0bf796b0
Closes-Bug: #1473535
Saw the error below in gerrit logs:
"Failed to connect to 2610:28:3090:3001:5054:ff:feb9:6ae0: Network is
unreachable" when trying to access fedorapeople.org.
Change-Id: I96603948560bc7ea259235cf3b304b2e64ceb5ab
Due to the licensing issues and the modules not existing in upstream
Ansible yet, I have written a simple module to fill the gaps.
This also uses Keystone v2.0 for all create of users, roles and
endpoints. The implementation of Keystone v3 must be discussed after the
new modules arrive.
Partially-Implements: blueprint ansible-service
Change-Id: I389edd56741360dd26fbbc0a982f365ca27ff446
AIO nodes had an address in the gcomm list which would prevent AIO
containers from starting appropriately.
Change-Id: I6db8c831ef9f9c0348748eb9539db326dc0df048
Closes-Bug: #1472481
The following files had requirement for root user
which can be removed as long as as the the user
executing script is part of docker group
Scripts updated
tools/kolla
tools/cleanup-containers
Closes-Bug: #1473551
Change-Id: Ia4aea205f5da20fb1ded4f7a5d11eeebe5f40aa9
The "hacking" package used in most OpenStack repos pulls in
additional pep8 rules. This commit should fix all of them
other than a hand full in the build.py file which will
be fixed in a separate commit
Change-Id: I40a0a30d338592ad64123e3da5523b9ad95a7a9c
Partially-Implements: blueprint enable-flake8
Just checking for an account in keystone is pointless since nova and
neutron take way more time before they are fully operational. If the user
would continue as tools/kolla suggests, the init-runonce script would
fail because either nova or neutron would not be ready for service.
Change-Id: I8fbfaa16d961b418b9ac6a34d311adb8e41b61e9
Closes-bug: #1471331
oslo.reports is needed by several services. I suspect more services
will need this package as time passes.
Change-Id: I306d03bb9fb62e2aed16113c1b547257a398cd1d
Closes-Bug: #1473232
