Age | Commit message (Collapse) | Author | Files | Lines |
|
Add container-specific variants of scenarios. These variants are
supposed to be temporary, as their only purpose is to allow us CIing the
scenarios with containers while we don't have pacemaker containerized
yet. Once we can deploy and upgrade containerized deployment with
pacemaker, these should be deleted and normal scenarios should be used.
Alternative approach would be to edit the scenarios on the fly within
the CI job, to remove the pacemaker parts, which would be more DRY, but
perhaps more surprising when trying to debug issues.
Change-Id: I36ef3f4edf83ed06a75bc82940152e60f9a0941f
|
|
|
|
|
|
proxy"
|
|
|
|
Use the openstack upper-constraints when running tox.
Change-Id: I9eef36eec749beec0effdb2309fe2ceb9bc557f8
Related-Bug: #1691511
|
|
|
|
|
|
|
|
The Apache certs were were being set even if TLS everywhere isn't
enabled. This fixes that.
Change-Id: If143d1fdeb0102a1c13441f89acaa73af24bf48f
|
|
This helps a bit with debugging issues, and the container will be
deleted on the next run when the same volume is configured.
Change-Id: I4f2f219bd7e40abafd0eb31c1275fdd8ed4db4da
|
|
|
|
|
|
|
|
Depends-on: I30ba93f76171e5993b5f0e1d7f1f5533acb25740
Closes-bug: #1668925
Change-Id: I3cb61d2d8765f9c2601bb00c4bfa24162883b96a
|
|
This spawns an extra container that runs httpd to run the TLS proxy that
will go in front of neutron server.
bp tls-via-certmonger-containers
Change-Id: I2529d78e889835f48c51e12d28ecd7c48739b02b
|
|
For TLS everywhere, neutron-server needs httpd in the image, since
it'll use a separate container that runs a TLS proxy to terminate
the connection. This requires the image where the configuration is
ran to have httpd installed, since there are several directories
and the user/group that's needed.
So, we then switch the image to be used to be neutron-server instead
of the openvswitch-agent image.
Change-Id: Ie16de3004925b7624f106d6c015ec04ef6031a06
Depends-On: I82f10ac0e7e692e6ba4a06dc10da9eaf79c60e7e
|
|
Changing the default values of neighbor table (also known as ARP table)
in the kernel to avoid neighbour table overflow and thus fix
communication errors between overcloud nodes.
default kernel values support L2 network up to 1024 hosts (/22).
The patch will allow up to 4096 hosts (/20).
Change-Id: I5fabc766dd75a38cd3d835deee7e168f04dd30ce
Closes-Bug: #1690087
|
|
The currently used resource type does not exist, therefore changing it.
Closes-Bug: 1691021
Change-Id: Iaf18af546817e0cf6cdfafcc5c54ab4d1a0f819d
|
|
This was forgotten in I72376a803ec6b2ed93903cc0c95a6ffce718b6dc and
broke containerized deployment.
Change-Id: I599a87bf06efbfefd3067c77ed6ca866505900f9
Closes-Bug: #1690870
|
|
|
|
All paths should be relative as we should not rely on the package
location - this can easily be overridden via --templates, and this
is exactly what we do for the upgrades job, where this will break
because we'll include the wrong (newer) version of these services
when deploying the older pre-upgrade overcloud.
Change-Id: Id8aea09305c0857253c44477945e34377cca64ca
|
|
|
|
Currently we just use what puppet-snmp provides in terms of defaults.
This means that currently every single snmp query gets logged with
the following:
May 15 10:51:30 centos.localdomain snmpd[5159]: Connection from UDP: [127.0.0.1]:57799->[127.0.0.1]:161
May 15 10:51:30 centos.localdomain snmpd[5159]: Connection from UDP: [127.0.0.1]:57799->[127.0.0.1]:161
May 15 10:51:32 centos.localdomain snmpd[5159]: Connection from UDP: [127.0.0.1]:50566->[127.0.0.1]:161
The reason is that we use '-LS0-6d' as the default content for
/etc/sysconfig/snmpd:
https://github.com/razorsedge/puppet-snmp/blob/master/manifests/params.pp#L322
This default means that we are logging from 0 (LOG_EMERG) to 6
(LOG_INFO). The above messages bring nothing in a default installation
and only spam the log files, so let's lower the upper log level to 5
(LOG_NOTICE) by default, so we properly do not see every single query in
the logs. We add an option so the operator can still configure the
desired log level via a Heat parameter.
Change-Id: I8d3dfdb4d549cd27131346fc477755ad72313449
|
|
|
|
Service template's parameter documentation has been update by
correcting few of the wrong informations and added more
information with examples.
Change-Id: I2d92fd01cbeb6fdc6f030255dc4b71166509b4f6
|
|
|
|
|
|
When a service is enabled on multiple roles, the parameters for the
service will be global. This change enables an option to provide
role specific parameter to services and other templates.
Two new parameters - RoleName and RoleParameters, are added to the
service template. RoleName provides the role name of on which the
current instance of the service is being applied on. RoleParameters
provides the list of parameters which are configured specific to the
role in the environment file, like below:
parameters_default:
# Default value for applied to all roles
NovaReservedHostMemory: 2048
ComputeDpdkParameters:
# Applied only to ComputeDpdk role
NovaReservedHostMemory: 4096
In above sample, the cluster contains 2 roles - Compute, ComputeDpdk.
The values of ComputeDpdkParameters will be passed on to the templates
as RoleParameters while creating the stack for ComputeDpdk role. The
parameter which supports role specific configuration, should find the
parameter first in in the RoleParameters list, if not found, then the
default (for all roles) should be used.
Implements: blueprint tripleo-derive-parameters
Change-Id: I72376a803ec6b2ed93903cc0c95a6ffce718b6dc
|
|
|
|
|
|
|
|
This spawns an extra container that runs httpd to run the TLS proxy that
will go in front of glance-api.
bp tls-via-certmonger-containers
Change-Id: If902ac732479832b9aa3e4a8d063b5be68a42a9b
|
|
This spawns an extra container that runs httpd to run the TLS proxy that
will go in front of swift.
bp tls-via-certmonger-containers
Depends-On: Ib01137cd0d98e6f5a3e49579c080ab18d8905b0d
Change-Id: I9639af8b46b8e865cc1fa7249bf1d8b1b978adfe
|
|
|
|
|
|
Change-Id: Ifaa3bb0400ee22601fd8f3e1f2f16192b5f8766b
|
|
We dont need expirer unless we have collector and standard
storage enabled. Lets turn it off by default and make it
an optional service. In upgrade scenario, we will kill the
process and stop the expirer, unless explicitly enabled.
Change-Id: Icffb7d1bb2cf7bd61026be7d2dcfbd70cd3bcbda
|
|
|
|
|
|
Once puppet has written the initial fernet keys, if a deployer wants to
rotate them, the keys will be overwritten when another overcloud deploy
is executed (for instance, for updates or upgrades). This disables
replacing this keys via puppet, so now the operator can rotate the keys
out of band.
Change-Id: I01fd46ba7c5e0db12524095dc9fe29e90cb0de57
|
|
|
|
Change-Id: I3583a9a3bb04df2aebf06a566a2bdc4afdbfc9f3
|
|
|
|
neutron-metadata number of workers will be taken from NeutronWorkers parameter
if not empty. when empty, all keys related to NeutronWorkers value will be
set with empty dictionary instead empty string ({}).
Change-Id: I18347639c188bbf085e2f3c739465e52c94b9d77
Closes-bug: #1689571
|
|
We need Docker service mapping defined and set to OS::Heat::None so that
we can reuse multinode-container-upgrade.yaml service list both for
initial deployment and for the upgrade. The upgrade will not be broken
by this as its env files are being passed later on the command line, and
they'll take priority and effectively enable the Docker service on
upgrade.
Another change we need for mixed upgrade is to add the TripleoPackages
service, which will take care of updating RPMs on the bare metal and
prevent docker installation from failing with outdated
puppet-tripleo ("Could not find class ::tripleo::profile::base::docker").
Related-Bug: #1685795
Closes-Bug: #1689772
Change-Id: Idb6917f22d0e9f74f8853972c6a08bffb01be410
|
|
|
|
Variables are now passed in with --env in the docker run call.
This will allow docker-puppet.sh to be baked into the image instead of
having it as a custom entrypoint.
Change-Id: Icbaefe033becc6b2226535f28ee202917bdc1074
|
|
Move the Zaqar WSGI service to use httpd in docker deployment.
Depends-On: I35cfd1c2320eb972890b44668c8f9f0a047a65dc
Change-Id: I56a6469a9179b5c023738f447e7665d0d3c73d0b
Co-Authored-By: Martin André <m.andre@redhat.com>
Co-Authored-By: Thomas Herve <therve@redhat.com>
|
|
Change I5c8b0c4abfc0607f42fd3f2da9f5ef2702b1bbe1 introduced conditions
to optimize upgrade times and fix related bugs. Unfortunately the
conditional inclusion would have to be paired with support in depends_on
to work as we need. Currently we can hit this bug if the batch upgrade
steps are undefined for some role, but upgrade steps are definied:
The specified reference "ControllerUpgradeBatch_Step2" (in
ControllerUpgradeConfig_Step0) is incorrect.
To fix this we have to make the steps unconditional. This isn't fully
reverting the original change because that change also addressed
ordering issues.
Change-Id: I369591f4757c10142f5b455e64aa778e1a9a5611
Closes-Bug: #1689553
|