Due to the recent Ceph 6 (Quincy) container image refactoring,
the old entrypoint is no longer available.
However, the ceph-nfs systemd unit should support both Pacific
and Quincy releases.
This patch introduces a task that is able to detect the version
provided by the ceph container in the Heat stack and renders the
systemd unit accordingly.
Change-Id: I34537db460ff5aaba93da7e104262e26e64cbb1a
The multi-cell stacks don't have networks and only tag
the ports. The code now simply mixes ports of multiple
stacks together. To resolve this we filter for the
tags of stacks in the vip ports.
Co-authored-by: Daniel Bengtsson <dbengt@redhat.com>
Change-Id: Id3eacdb631a7a7102d28ca4b9963fed3d99f2414
Some containers such as keystone_bootstrap contains secret information
in the container definitions. We should not dump such details into
ansible output.
Closes-Bug: #1998181
Change-Id: I6c98afc8e0d6822199001ffd9535575a4506b82d
During `openstack overcloud deploy`, ceph create pool operation
fails on RHSCSv6 as the podman command generated by 'ceph_pool'
ansible module has the volume argument as /etc/ceph.
This patch ensures to use the '/var/lib/ceph/$FSID/config/'
in the podman command generated by all the ansible moduels.
Even the 'tripleo_cephadm_ceph_cli' is updated to use the
new path.
Resolves: rhbz#2172582
Change-Id: I570ce859c3033d0f8cc36064f851419fca1a9d66
The lb-mgmt-subnet is added to a router but Neutron requires that the
subnet has a non-empty gateway.
Ensure it exists before adding it.
wallaby-backport-potential
Change-Id: I1e01880449bd2fc08da949dc409e4089fe2f6e0d
If selinux is disabled, we don't need to, and can't set the fcontext on
the list of files. This change adds a conditional statement to this task
based on the status of selinux.
Closes-Bug: #2008822
Change-Id: Id0135cc2456b1716e3bedace16ca4b9cee89d7b4
Ensure iscsid.conf exists, specifically to cover the case where the
iscsi-initiator-utils package isn't installed on the host.
Closes-Bug: #2007839
Change-Id: Ic057183fcdd208295777492e1836f0757cc9d579
Existing implementation could lead to module failing with exception
if defaults were not provided.
Resolves: rhbz#2136519
Signed-off-by: Jiri Podivin <jpodivin@redhat.com>
Change-Id: I02d2452347b0ab6d7f79d2756b5f5dbc5f0cef84
Process based locking was added to improve container prepare
performance for python2/mistral, according to this change[1] so it
might not be appropriate to use it in an ansible module.
This change removes the ProcessLock so the image preparation uses the
default ThreadLock (and the thread executor).
It is possible that authentication caching doesn't work at all with
the process based executor, and support for that executor should be
removed entirely now that it is not required.
[1]60afc0eec4
Resolves: rhbz#2165527
Change-Id: Ib127b11cc26944dc4c348d7c8b6938bbea0a134a
This patch adds configure task which replaces the iscsid configuration
in [1] with these modifications.
- configure /etc/iscsi on the host directly, so it eliminates the use
of /var/lib/config-data/ansible-generated
- remove "sync from host" and "sync to host" operations which are no
longer needed.
- optimize the install task.
[1] https://github.com/openstack/puppet-tripleo/blob/master/manifests
/profile/base/iscsid.pp
Co-Authored-By: Manoj Katari <mkatari@redhat.com>
Change-Id: Idb57bb5179897ee7b4461f81372e6021b72be4d8
Ansible package changed name to ansible-core. Exclude must stay due to
bz https://bugs.launchpad.net/tripleo/+bug/1998501, only update exclude
field to exclude ansible-core instead of ansible.
As a part of standalone-roles work update tasks from tht
deployment/tripleo-packages/tripleo-packages-baremetal-puppet.yaml were
migrated to tripleo-ansible and this is the reason why need tht
change[1] together with this tripleo-ansible change.
Related-Bug: #1998501
[1]https://review.opendev.org/c/openstack/tripleo-heat-templates/+/872568
Change-Id: I09586601036370696f0b93e7b171a0bcc5393b27
We had expected that octavia-amphora-image was always installed when
deploying Octavia (it was pulled by the rhos-director-images RPM) but
when deploying an UEFI overcloud, the package is not installed.
This commits adds a task that ensures the octavia-amphora-image package
is installed if:
- the user doesn't specify another image path
- the RPM is available (the RPM doesn't exist in RDO - tripleo CI jobs
download the image in an other ansible playbook)
wallaby-backport-potential
Resolves: rhbz#2132889
Change-Id: Ic5fafe54f84fd934839c340b2f7bb5455d8b8639
Fix the tripleo_multipathd role's task that detects whether a
blacklist section is present in /etc/multipath.conf. Previously,
an error caused it to insert malformed lines in the file.
Resolves: rhbz#2159555
Change-Id: If85e8c3644af21455ce497d88ee975c056f4fcaa
Downstream Wallaby backport candidate.
This will use the existing lb_mgmt_subnet_cidr setting
to decide whether IPv6 is used and configure the deployment accordingly.
Change-Id: I735a905fa42a354cb2547e4fc96c396af914a529
When executing the overcloud backup command, the db backup files are missing since the when condition for executing the db_backup.yaml file is never met.
Change-Id: I0eed37d1da74acf9d7910c6ed0c2ac47adadda12
01-site:
Ensure we get a consistent format (yaml)
02-computes + host_vars/localhost:
Ensure the "ansible_connection" parameter is passed from within the
inventory, so that it's more obvious. Until now, we may just edit the
02-computes to inject our remote hosts, and not get what we want since
ansible will just use the local node instead of connecting to the
remotes.
Change-Id: I33b07d615beafecd5fe97cbca54c6a296965af8f