Kolla disables those repos in https://review.opendev.org/#/c/725032/
patch. This breaks TripleO builds as they assume them to be enabled.
Change-Id: Id6ea8f2a73c5205c3b3626bac42a55a0763feef4
Since validations don't rely on mistral anymore, it's time to remove
some useless package and user.
Change-Id: Ibcee290b4eb3ec3e5ee7af747a9ca7242863147f
Since centos-release-opstools repo is now available for centos-7
and centos-8. So we do not need distro conditional for that.
Change-Id: I97e2bcd55f4bff448f9c492d8287ac08832010b2
Signed-off-by: Chandan Kumar (raukadah) <chkumar@redhat.com>
Starting with the ussuri release, python3-networking-mlnx is
added to the neutron-base container Dockerfile.
Since python3-networking-mlnx is not supported in all cases,
this patch adds an override to remove it when the container
is built.
Change-Id: I37497509ec7a61f815247c34c0703a75f18f0b6e
Related-Bug: BZ#1828353
In order to support configuration of ed25519 authentication,
puppet-tripleo uses a python helper that depends on pynacl.
Update the kolla config override to install pynacl in the mariadb
image to support the ed25519 use case.
Change-Id: I908b2af6acce25dfb115463f9d04b0e14c7cba33
Related-Bug: #1866093
In placement, mysql-migrate-db.sh calls the mysql CLI directly,
and when users are configured to authenticate via ed25519, mysql
CLI requires client_ed25519.so, which is provided by mariadb-server.
Partial-Bug: #1866093
Change-Id: I011d0056c34150420eb4b9dfde721a7f30a0e25a
To reuse the tripleo containers when running libvirtd container
on OCP we need to run it via a wrapper script and create our own
cgroups and run libvirtd using cgexec to break out the default
OCP cgroup of the pod. Otherwise if the libvirtd pod gets
deleted/recreated all VMs get shutdown as OCP cleans up the
resources owned by the pod.
Change-Id: I469b4f51c57e1651f42064b0134aa3916af72d80
Following changes are needed for building CentOS8
containers:-
- centos-release-opstools repo is not available for
CentOS8 yet, until it get's available it need not
run in CentOS8.
- Highavailability repo need to be enabled for CentOS8
to setup HA packages.
- Also add rabbitmq-server-3.7.* ro rabbitmq_packages
remove list.
- remove influxdb from repo disable as it's not enabled
for CentOS8.
Change-Id: I4c0c27f54717a2701388c138bbe40ffc0308e5f1
Kolla will upgrade rabbitmq version to 3.8 for Ussuri, so need to update
tripleo template overrides. This adds the new pin, which will be cleaned
up on a follow on review.
Implements: blueprint rabbitmq-version-upgrade
Change-Id: I8a08b2e705e35d880f5fdacedf506009f00ac2fa
Closes-Bug: #1853801
The health check has existed for some time but needs to be included in
the container.
Change-Id: Ia8a9517a3e505ad47266b90bbc90425b119a4ea5
Related-Bug: #1856088
Kolla upstream is adding centos-release-openstack-train release rpm in
master branch after [1]. In tripleo builds this means it's configuring
both RDO Trunk master (delorean) repos and CloudSIG train repo. Mixing
both sets of repos can lead to undesired results so I'm removing
centos-release-openstack-train repo from config and letting
tripleo-repos configure as needed.
Also, I'm removing leftovers from old Ceph releases configuration.
Currently kolla is installing centos-release-ceph-nautilus so, tripleo
does not need to overrides it.
[1] https://review.opendev.org/#/c/689706
Change-Id: Ia551a7d2351bb91cc45e9c2d12f4afb4f6368bb7
We commented this out back in late 2017 since they reogranized the
containers. Since no one has picked up this effort, let's drop this dead
code.
Change-Id: I45f0787e92eca59ab7eec09b2b199c5b1dbc4855
This change organizes the override template file into
header/packages/footer related sections and alphabetizes each of these.
Change-Id: I96734bf9381a5cff0b9af4aed6a25b9b2da412b6
This is required in a nova-less scenario where the provision workflow
needs to include the novajoin cloud-init script in the config_drive
data so that the provisioned node can be registered with FreeIPA.
Change-Id: Ie57ab22b72b172985a3fff02bb4d8915c9df4152
Blueprint: nova-less-deploy
This change will add an empty block for `fluentd_monasca_plugin_install` to
ensure there are no plugins installed. In a proposed change in kolla, a new
block is being added specifically for the monasca plugins, which is causing
our container build process to fail in their external CI. By blocking this
out we'll maintain our expected stance of not installing any plugins.
Change-Id: Ic7f2d12073883bce4e03a48ff6a2515acf6298bf
Related-Change: https://review.opendev.org/675614
Signed-off-by: Kevin Carter <kecarter@redhat.com>
Kolla is adding the td-agent package to their fluentd container. This
package is provided by a third-party and not something we want to
support at this time.
Depends-On: https://review.opendev.org/#/c/677063
Change-Id: I3a5a5ed0defe8ce0259e5c09eaec11814fd51054
Signed-off-by: Kevin Carter <kecarter@redhat.com>
This patch adds systemd-udev package. This package is not contained
on RHEL8 images by default which causes following error when the agent
is verifying container health from within the container:
"Failed to connect to bus: No data available"
More info: https://bugzilla.redhat.com/show_bug.cgi?id=1728226
Change-Id: I852bd75ab4a6fb8b48fd68ab70e804ec8e3c942d
https://review.opendev.org/#/c/672547 adds collectd-sensubility
to the collectd image but it is not available in RHEL-8
leading to blocking the promotion pipeline.
In order to fix it, we are removing it from collectd
container till it is available.
Related-Bug: #1839423
Change-Id: Ie01c20a00287e0a58277db664dd0a278d317d814
Signed-off-by: Chandan Kumar (raukadah) <chkumar@redhat.com>
The nova_libvirt containers need to be able to launch containers for
the virtlogd process to be not managed by paunch.
We only add docker if distro_python_version.startswith('2') . This
applies to
- nova_libvirt_packages_append
- neutron_metadata_agent_ovn_packages_append
- neutron_l3_agent_packages_append
- neutron_dhcp_agent_packages
Related-Bug: #1838272
Change-Id: I5829addb715c4836dd27750daa67f2e472084b34
In review of the usage within openstack, it doesn't appear to actually
be in use by any actual services. There is a few references in designate
(for a contrib file) and networking-cisco. Since this doesn't affect (or
is used by tripleo), let's drop this dependency.
Change-Id: Ia406cc91607d6612b60c61ab0636ddecf8c89061
As of Rocky [1], the nova-consoleauth service has been deprecated and
cell databases are used for storing token authorizations. All new consoles
will be supported by the database backend and existing consoles will be
reset. Console proxies must be run per cell because the new console token
authorizations are stored in cell databases.
nova-consoleauth was deprecated in tripleo with:
I68485a6c4da4476d07ec0ab5e7b5a4c528820a4f
This change now removes the NovaConsoleauth Service.
[1] https://docs.openstack.org/releasenotes/nova/rocky.html
Depends-On: https://review.opendev.org/658081
Change-Id: I21012e25a8f320dcea11d51f9a8cfa5d76520e78
Related-Bug: #1828414
From the OpenStack Stein release we want to deploy the Ceph Nautilus
version; this change updates the default container image to use accoringly.
Depends-On: Id0de75e7459b9a8d993a87552b51cf6fdbd29687
Change-Id: I133cd0cd632e0ba6b5a844a89bf74a063cf50f07
Context: https://bugzilla.redhat.com/show_bug.cgi?id=1693752
We agreed that we need dumb-init upstream and downstream,
one of the reasons is to handle zombie processes in the
containers.
e.g. ssh connections in mistral_executor container,
created by Ansible playbook runs.
This reverts commit 89cbba272d.
Related-Bug: 1821854
Change-Id: I93f97f76b598da268a77ff356fb1bffddfddef5f
The dumb-init is actually installed by fetching it from github and
putting it in the base image. We don't actually want that for our usage.
Additionally a new block that overrode the entrypoint in the mariadb
container which now causes the mysql container to fail if dumb-init is
not installed.
Change-Id: I2228a0da506daf207ea402d3e5a0f156b629862a
Closes-Bug: #1821970
Emilien Macchi
yatinkarel