apex-tripleo-heat-templates.git
8 years agoFixup the memcached servers string in nova.conf for v6
marios [Wed, 20 Jan 2016 10:34:07 +0000 (12:34 +0200)]
Fixup the memcached servers string in nova.conf for v6

As discussed at https://bugzilla.redhat.com/show_bug.cgi?id=1299265
when providing a list of IPv6 addresses as the memcache_node_ips
the resulting nova.conf entry can't be parsed properly.

This adds a memcache_node_ips_v6 which has the required format like
inet6:[ADDR1],inet6:[ADDR2],inet6:[ADDR3]

Closes-Bug: 1536103
Change-Id: I7f95fa063cbba279c4c2e270841f0a279d2be2f6

8 years agoMerge "Add IPv6 Support to Isolated Networks"
Jenkins [Sat, 5 Mar 2016 02:20:16 +0000 (02:20 +0000)]
Merge "Add IPv6 Support to Isolated Networks"

8 years agoMerge "Allow for usage of pre-allocated IPs for the management network"
Jenkins [Fri, 4 Mar 2016 19:29:53 +0000 (19:29 +0000)]
Merge "Allow for usage of pre-allocated IPs for the management network"

8 years agoMerge "Run keystone-manage bootstrap for HA deployment too"
Jenkins [Fri, 4 Mar 2016 17:57:21 +0000 (17:57 +0000)]
Merge "Run keystone-manage bootstrap for HA deployment too"

8 years agoRun keystone-manage bootstrap for HA deployment too
Jiri Stransky [Fri, 4 Mar 2016 13:54:51 +0000 (14:54 +0100)]
Run keystone-manage bootstrap for HA deployment too

This is necessary to keep creating the Default domain.

Change-Id: Ib9911819e89f30270d4f7597639b33f30ad2e3a6
Closes-Bug: #1549867

8 years agoMerge "Set notification driver for nova to send"
Jenkins [Fri, 4 Mar 2016 14:13:05 +0000 (14:13 +0000)]
Merge "Set notification driver for nova to send"

8 years agoAdd IPv6 Support to Isolated Networks
Dan Sneddon [Thu, 15 Oct 2015 15:10:44 +0000 (08:10 -0700)]
Add IPv6 Support to Isolated Networks

This change adds a new set of network templates with IPv6 subnets
that can be used instead of the existing IPv4 networks. Each network
can use either the IPv4 or IPv6 template, and the Neutron subnet will
be created with the specified IP version.

The default addresses used for the IPv6 networks use the fd00::/8
prefix for the internal isolated networks (this range is reserved
for private use similar to 10.0.0.0/8), and 2001:db8:fd00:1000::/64
is used as an example default for the External network
(2001:db8::/32 are the documentation addresses [RFC3849]), but this
would ordinarily be a globally addressable subnet. These
parameters may be overridden in an environment file.

This change will require updates to the OpenStack Puppet
Modules to support IPv6 addresses in some of the hieradata values.
Many of the OPM modules already have IPv6 support to support IPv6
deployments in Packstack, but some OPM packages that apply only to
Instack/TripleO deployments need to be updated.

IPv6 addresses used in URLs need to be surrounded by brackets in
order to differentiate IP address from port number. This change
adds a new output to the network/ports resources for
ip_address_uri, which is an IP address with brackets in the case
of IPv6, and a raw IP address without brackets for IPv4 ports.
This change also updates some URLs which are constructed in Heat.

This has been tested and problems were found with Puppet not
accepting IPv6 addresses. This is addressed in the latest Puppet.
Additional changes were required to make this work with Ceph.

IPv6 tunnel endpoints with Open vSwitch are not yet supported
(although support is coming soon), so this review leaves the
Tenant network as an isolated IPv4 network for the time being.

Change-Id: Ie7a742bdf1db533edda2998a53d28528f80ef8e2

8 years agoAllow for usage of pre-allocated IPs for the management network
Steven Hardy [Thu, 11 Feb 2016 10:57:54 +0000 (10:57 +0000)]
Allow for usage of pre-allocated IPs for the management network

Id3d4f12235501ae77200430a2dc022f378dce336 added support for pre-allocated
IPs on the other overlay networks, but because the patch adding the
managment network (I0813a13f60a4f797be04b34258a2cffa9ea7e84f) was
under review around the same time, we missed adding the from_pool
capability to the ManagementNetwork.

Change-Id: If99f37634d5da7e7fb7cfc31232e926bd5ff074a

8 years agoMerge "Deploy Aodh services, replacing Ceilometer Alarm"
Jenkins [Thu, 3 Mar 2016 20:40:03 +0000 (20:40 +0000)]
Merge "Deploy Aodh services, replacing Ceilometer Alarm"

8 years agoDeploy Aodh services, replacing Ceilometer Alarm
Emilien Macchi [Tue, 3 Nov 2015 23:09:34 +0000 (18:09 -0500)]
Deploy Aodh services, replacing Ceilometer Alarm

Ceilometer Alarm is deprecated in Liberty by Aodh.

This patch:
* manage Aodh Keystone resources
* deploy Aodh API under WSGI, Notifier, Listener and Evaluator
* manage new parameters to customize Aodh deployment
* uses ceilometer DB for the upgrade path
* pacemaker config

Depends-On: I9e34485285829884d9c954b804e3bdd5d6e31635
Depends-On: I891985da9248a88c6ce2df1dd186881f582605ee
Depends-On: Ied8ba5985f43a5c5b3be5b35a091aef6ed86572f

Co-Authored-By: Pradeep Kilambi <pkilambi@redhat.com>
Change-Id: I58d419173e80d2462accf7324c987c71420fd5f6

8 years agoMerge "endpoint_map: bump nova api from 2.0 to 2.1"
Jenkins [Thu, 3 Mar 2016 15:33:04 +0000 (15:33 +0000)]
Merge "endpoint_map: bump nova api from 2.0 to 2.1"

8 years agoMerge "Use set -e for validation-scripts/all-nodes.sh"
Jenkins [Thu, 3 Mar 2016 15:12:13 +0000 (15:12 +0000)]
Merge "Use set -e for validation-scripts/all-nodes.sh"

8 years agoMerge "Moves the swift start/stop into the common_functions.sh file"
Jenkins [Thu, 3 Mar 2016 12:48:53 +0000 (12:48 +0000)]
Merge "Moves the swift start/stop into the common_functions.sh file"

8 years agoMerge "Add Satellite 5 support"
Jenkins [Thu, 3 Mar 2016 12:00:31 +0000 (12:00 +0000)]
Merge "Add Satellite 5 support"

8 years agoMerge "Cisco nexus config template - obsolete parameter (replay count)."
Jenkins [Thu, 3 Mar 2016 09:51:24 +0000 (09:51 +0000)]
Merge "Cisco nexus config template - obsolete parameter (replay count)."

8 years agoMoves the swift start/stop into the common_functions.sh file
marios [Wed, 2 Mar 2016 16:31:51 +0000 (18:31 +0200)]
Moves the swift start/stop into the common_functions.sh file

Since swift isn't managed by pacemaker we need to manually (systemctl)
stop and start the swift services. This moves the duplicate blocks for
start/stop into a common function (we already include that
pacemaker_common_functions.sh here so may as well)

Change-Id: Ic4f23212594c1bf9edc39143bf60c7f6d648fd1d

8 years agoMerge "Upgrades: install zaqarclient"
Jenkins [Wed, 2 Mar 2016 15:39:24 +0000 (15:39 +0000)]
Merge "Upgrades: install zaqarclient"

8 years agoMerge "Upgrades: quiet yum update"
Jenkins [Wed, 2 Mar 2016 15:32:28 +0000 (15:32 +0000)]
Merge "Upgrades: quiet yum update"

8 years agoMerge "Support adding a swap file to overcloud nodes"
Jenkins [Wed, 2 Mar 2016 12:14:39 +0000 (12:14 +0000)]
Merge "Support adding a swap file to overcloud nodes"

8 years agoUpgrades: install zaqarclient
Jiri Stransky [Mon, 29 Feb 2016 16:44:37 +0000 (17:44 +0100)]
Upgrades: install zaqarclient

Old overcloud images don't have python-zaqarclient installed, and new
overclouds' os-collect-config are configured with Zaqar support. This
together means that on upgrade we need to install python-zaqarclient,
otherwise os-collect-config will be restarted during yum update and
crash due to trying to import missing Python module from zaqarclient.

Change-Id: I3e875e14cb60b1b78aec0d9ddc412ccf865abd01

8 years agoUpgrades: quiet yum update
Jiri Stransky [Mon, 29 Feb 2016 16:38:42 +0000 (17:38 +0100)]
Upgrades: quiet yum update

Quiet down yum during major upgrades to reduce the output size. This is
consistent with what was introduced into minor updates in change
I517271e8465885421a78b73c5af756816c37a977.

Change-Id: Ie6b470e383fdf42870ac6f60ca43e44b4c446ebe

8 years agoMerge "Use service tenant for ceilometer"
Jenkins [Wed, 2 Mar 2016 08:48:31 +0000 (08:48 +0000)]
Merge "Use service tenant for ceilometer"

8 years agoMerge "Add HostnameMap to allow granular control of hostnames"
Jenkins [Wed, 2 Mar 2016 07:15:22 +0000 (07:15 +0000)]
Merge "Add HostnameMap to allow granular control of hostnames"

8 years agoSupport adding a swap file to overcloud nodes
James Slagle [Thu, 28 Jan 2016 20:34:54 +0000 (15:34 -0500)]
Support adding a swap file to overcloud nodes

Create a new SoftwareDeployment that can be used to add a swap file to
all nodes The amount of swap and the location of the swap file can be
customized via parameter_defaults and the swap_size_megabytes/swap_path
parameters.

Change-Id: I1fb14c0fab2255410fceb26c3a7d5cfe0ba57b3b

8 years agoMerge "Enable heat-manage purge_deleted cron job"
Jenkins [Tue, 1 Mar 2016 18:03:35 +0000 (18:03 +0000)]
Merge "Enable heat-manage purge_deleted cron job"

8 years agoMerge "Update the path to the Docker file"
Jenkins [Tue, 1 Mar 2016 18:01:24 +0000 (18:01 +0000)]
Merge "Update the path to the Docker file"

8 years agoendpoint_map: bump nova api from 2.0 to 2.1
Emilien Macchi [Tue, 1 Mar 2016 16:43:24 +0000 (11:43 -0500)]
endpoint_map: bump nova api from 2.0 to 2.1

Nova v2.1 allows to use the same API as 2.0 but with microversions
support, which is the recommended way to discover the latest API
version supported in the cloud.

Change-Id: Id011de03d883001fd48dbbcfed53cb821607c7f3

8 years agocontroller/ha: disable keystone-manage bootstrap.
Emilien Macchi [Tue, 1 Mar 2016 01:48:27 +0000 (20:48 -0500)]
controller/ha: disable keystone-manage bootstrap.

Because Overcloud Keystone resources are not managed by puppet-keystone
but by os-cloud-config, we need to let os-cloud-config managing keystone
bootstrap otherwise the Exec will fail since some data is already in
place.

Later, when Keystone resources will be managed by Puppet, drop this
parameter, because puppet-keystone is able to manage the boostrap
itself.

Change-Id: I027deaae5cf90c27a6b5e9d236ae61145cab3c3f
Closes-Bug: #1551501

8 years agoMerge "Convert port cidr splitting to str_split"
Jenkins [Mon, 29 Feb 2016 20:56:04 +0000 (20:56 +0000)]
Merge "Convert port cidr splitting to str_split"

8 years agoCisco nexus config template - obsolete parameter (replay count).
Leon Zachery [Mon, 29 Feb 2016 19:06:44 +0000 (14:06 -0500)]
Cisco nexus config template - obsolete parameter (replay count).

Due to fix: https://bugs.launchpad.net/networking-cisco/+bug/1469839,
the replay count parameter is no longer used.  This needs to be
reflected in the Triple O templates.

Change-Id: I666c4394108287adcb4989e897ab3936667a602b
Closes-bug:  #1551387

8 years agoAdd Satellite 5 support
James Slagle [Tue, 12 Jan 2016 22:34:22 +0000 (17:34 -0500)]
Add Satellite 5 support

Add Satellite 5 support to the RHEL registration environment and
resources. The registration script is updated to support both satellite
versions in place given the similarity of the options for both
scenarios.

The satellite version is detected based on $REG_SAT_URL, and that
determines whether subscription-manager or rhnreg_ks is used.

Change-Id: Ic261c8a16a7d6d3978f8bfc6e53f75dbe1b716db

8 years agoMerge "OpenContrail heat templates"
Jenkins [Mon, 29 Feb 2016 15:31:51 +0000 (15:31 +0000)]
Merge "OpenContrail heat templates"

8 years agoMerge "Change the default value for NetworkNexusVxlanGlobalConfig"
Jenkins [Mon, 29 Feb 2016 15:11:48 +0000 (15:11 +0000)]
Merge "Change the default value for NetworkNexusVxlanGlobalConfig"

8 years agoUse set -e for validation-scripts/all-nodes.sh
Dan Prince [Mon, 29 Feb 2016 01:19:49 +0000 (20:19 -0500)]
Use set -e for validation-scripts/all-nodes.sh

Update this script to use 'set -e' for all commands except
the ping checks themselves which we allow to fail so we
can give enhanced output.

This resolves an issue where some commands could fail and cause
an undetectable error (the scripted exits with success) thus causing
a case where Heat won't detect any network errors at all. This
was recently the case with git commit 45be848 where we switched
to use python -c 'import ipaddr' which wasn't even installed as a
library on our overcloud nodes, thus causing all network validations
to silently fail.

Change-Id: I40ed6a537136e866357cc0d9304e905afdd28522
Depends-On: Ia617f44b4673b89202e5e5cdcac9b50f46b3e6c8
Related-bug: #1551048

8 years agoMerge "Write the compute upgrade script for tripleo major upgrade workflow"
Jenkins [Mon, 29 Feb 2016 10:03:15 +0000 (10:03 +0000)]
Merge "Write the compute upgrade script for tripleo major upgrade workflow"

8 years agoMerge "Enable notifications on the overcloud"
Jenkins [Mon, 29 Feb 2016 09:29:27 +0000 (09:29 +0000)]
Merge "Enable notifications on the overcloud"

8 years agoMerge "Add support for DeployArtifactURLs"
Jenkins [Sat, 27 Feb 2016 00:20:48 +0000 (00:20 +0000)]
Merge "Add support for DeployArtifactURLs"

8 years agoMerge "Add NovaVNCProxy back into endpoint_map"
Jenkins [Fri, 26 Feb 2016 23:37:23 +0000 (23:37 +0000)]
Merge "Add NovaVNCProxy back into endpoint_map"

8 years agoMerge "Emits a different hostname for each network the node is on"
Jenkins [Fri, 26 Feb 2016 23:28:25 +0000 (23:28 +0000)]
Merge "Emits a different hostname for each network the node is on"

8 years agoAdd support for DeployArtifactURLs
Dan Prince [Fri, 13 Nov 2015 20:01:13 +0000 (15:01 -0500)]
Add support for DeployArtifactURLs

Adds a new nested stack deployment which allows operators to
opt-in to deploy tarball's and RPM packages by setting
DeployArtifactURLs as a parameter_default in a Heat
environment.

The intent is to use this setting to allow t-h-t to
transparently deploy things like tarballs of puppet modules
via a Swift Temp URL.

Change-Id: I1bad4a4a79cf297f5b6e439e0657269738b5f326
Implements: blueprint puppet-modules-deployment-via-swift

8 years agoMerge "Nova RPC unpinning"
Jenkins [Fri, 26 Feb 2016 15:40:55 +0000 (15:40 +0000)]
Merge "Nova RPC unpinning"

8 years agoMerge "Add meta notify=true to rabbitmq resource"
Jenkins [Fri, 26 Feb 2016 10:29:30 +0000 (10:29 +0000)]
Merge "Add meta notify=true to rabbitmq resource"

8 years agoWrite the compute upgrade script for tripleo major upgrade workflow
marios [Mon, 22 Feb 2016 15:18:16 +0000 (17:18 +0200)]
Write the compute upgrade script for tripleo major upgrade workflow

As part of the major upgrade workflow non-controller nodes are to
be updated by the operator, out-of-band and only after an initial
heat stack-update that invokes the upgrade of the controller nodes.

This review adds a ComputeDeliverUpgradeConfigDeployment_Step3
SoftwareDeploymentGroup to be applied only to compute nodes, and
that depends on the controllers having been upgraded after
ControllerPacemakerUpgradeConfig_Step2.

Its purpose is to deliver but not invoke the upgrade script on
compute nodes to /root/tripleo_upgrade_node.sh .

The non-controller nodes will then be upgraded later by an
operator that will run the script provided for that purpose, like
at https://review.openstack.org/#/c/284722/1 for example.

Change-Id: Ic6115fc8cf5320abfcf500112ff563bde8b88661

8 years agoMerge "Add a sample network-environment.yaml file to environments"
Jenkins [Fri, 26 Feb 2016 09:58:54 +0000 (09:58 +0000)]
Merge "Add a sample network-environment.yaml file to environments"

8 years agoAdd NovaVNCProxy back into endpoint_map
Zane Bitter [Tue, 2 Feb 2016 17:32:37 +0000 (12:32 -0500)]
Add NovaVNCProxy back into endpoint_map

Due to an incorrect rebase, d0dcb9401c868786df58f5801a431392b8e89df8
dropped the changes made in dd7602ad82100617126be26d80a6d3f67cb739ac to
add a vncproxy to the endpoint map. This change restores them.

Change-Id: Ifef7f955481405d5fe39ba48c8b1a79aa9c170f2

8 years agoSet notification driver for nova to send
Pradeep Kilambi [Tue, 23 Feb 2016 16:45:57 +0000 (11:45 -0500)]
Set notification driver for nova to send

Currently since nova compute is not configured to
send notifications to ceilometer, tempest tests
fail on tempest.api.telemetry.test_telemetry_notification_api.

Change-Id: I763b7d246ae3f5955b6f555c8fd107d2cac89787

8 years agoEnable notifications on the overcloud
Ben Nemec [Wed, 4 Nov 2015 00:40:36 +0000 (18:40 -0600)]
Enable notifications on the overcloud

Configures all services to send notifications to rabbit.  The puppet
modules are not consistent regarding how this is done - some expose
notification config as a top-level param, others you need to set it
through a *_config structure, and cinder provides a separate class
dedicated to enabling ceilometer notifications.

Change-Id: I23e2ddad3c59a06cfbfe5d896a16e6bad2abd943

8 years agoMerge "Add UpgradeLevelNovaCompute parameter"
Jenkins [Thu, 25 Feb 2016 12:15:16 +0000 (12:15 +0000)]
Merge "Add UpgradeLevelNovaCompute parameter"

8 years agoMerge "Introduce update/upgrade workflow"
Jenkins [Thu, 25 Feb 2016 11:00:49 +0000 (11:00 +0000)]
Merge "Introduce update/upgrade workflow"

8 years agoMerge "Add resources for major upgrade in Pacemaker scenario"
Jenkins [Thu, 25 Feb 2016 11:00:41 +0000 (11:00 +0000)]
Merge "Add resources for major upgrade in Pacemaker scenario"

8 years agoMerge "Generate the endpoint map statically"
Jenkins [Thu, 25 Feb 2016 10:09:36 +0000 (10:09 +0000)]
Merge "Generate the endpoint map statically"

8 years agoAdd a sample network-environment.yaml file to environments
Dan Sneddon [Tue, 23 Feb 2016 20:51:40 +0000 (12:51 -0800)]
Add a sample network-environment.yaml file to environments

This change adds a sample network-environment.yaml file to the
environments. This sample includes pointers to NIC config files,
as well as default network subnets and allocation pools.

This is meant to be a demonstration of the default settings for
a virtual deployment. In a real deployment, the operator would
customize the settings here and point to custom NIC config
templates.

Change-Id: I0288c0680effea06b5f805a0d955e8bbf6152ba6

8 years agoMerge "Rename tox env to pep8"
Jenkins [Wed, 24 Feb 2016 22:25:30 +0000 (22:25 +0000)]
Merge "Rename tox env to pep8"

8 years agoMerge "Adds v6 capability to the deploy validation test (pings)"
Jenkins [Wed, 24 Feb 2016 18:46:05 +0000 (18:46 +0000)]
Merge "Adds v6 capability to the deploy validation test (pings)"

8 years agoEmits a different hostname for each network the node is on
Giulio Fidente [Wed, 20 Jan 2016 13:36:55 +0000 (14:36 +0100)]
Emits a different hostname for each network the node is on

Populates /etc/hosts with an entry for each IP address the node
is on, which will be useful to migrate services configuration from
using IPs into using hostnames.

This is how the lines look like on a host which doesn't have all ports:

  172.16.2.6 overcloud-novacompute-0.localdomain overcloud-novacompute-0
  192.0.2.9 overcloud-novacompute-0-external
  172.16.2.6 overcloud-novacompute-0-internalapi
  172.16.1.6 overcloud-novacompute-0-storage
  192.0.2.9 overcloud-novacompute-0-storagemgmt
  172.16.0.4 overcloud-novacompute-0-tenant
  192.0.2.9 overcloud-novacompute-0-management

the network against which the default (or primary) name is resolved
can be configured (for computes) via ComputeHostnameResolveNetwork

Change-Id: Id480207c68e5d68967d67e2091cd081c17ab5dd7

8 years agoNova RPC unpinning
Jiri Stransky [Wed, 24 Feb 2016 16:29:50 +0000 (17:29 +0100)]
Nova RPC unpinning

During upgrades, we only run Puppet on the whole deployment to converge
the state, after the upgrade workflow itself has been fully
completed. That is an opportunity to utilize Puppet to make sure Nova
Compute RPC doesn't remain pinned to the older version.

Change-Id: I6ebc813a80dfd9dfbbb213c38724487e044507b8

8 years agoGenerate the endpoint map statically
Zane Bitter [Tue, 2 Feb 2016 17:32:37 +0000 (12:32 -0500)]
Generate the endpoint map statically

A stack is an extremely heavyweight abstraction in Heat. Particularly in
TripleO, every stack includes a copy of all the template and environment
data for all of the stacks in the tree, all of which must be stored anew
in the database.

The EndpointMap abstraction created no fewer than 30 nested stacks, none
of which contained any resources but which existed purely for the
purpose of abstracting out some intrinsic functions used to calculate
the endpoint URLs for the various services. This likely adds several GB
to the memory requirements of the undercloud, and can cause things to
slow to a crawl since all 30 nested stacks need to be queried whenever
we need data from any one of them.

This change eliminates the nested stacks and instead generates the
endpoint map statically. This can be done offline in less than 250ms,
allows the input data to be expressed in an even more human-readable
form, and reduces the runtime overhead of the endpoints map by a factor
of 31, all with no loss of functionality, compatibility or flexibility.

Since we don't run a setup script to generate the tarball, the
endpoint_map.yaml output is checked in to source control. The build
script offers a --check option that can be used to make sure that the
output file is up-to-date with the input data.

Change-Id: I2df8f5569d81c1bde417ff5b12b06b7f1e19c336

8 years agoChange the default value for NetworkNexusVxlanGlobalConfig
Eugene Bagdasaryan [Thu, 18 Feb 2016 16:51:11 +0000 (19:51 +0300)]
Change the default value for NetworkNexusVxlanGlobalConfig

This parameter leads to the nonoperational state
in Nexus Vxlan topology when set to True, when VNIs created
but the nve peers do not get discovered on the Nexus.
It is time consuming process to debug the configuration
and find out that this parameter should be changed
to False. To prevent future problems for the future
deployment we want to default this parameter to False.

Change-Id: I685ad7d212af0d9e568acbf1ccf1607d120c195e

8 years agoMerge "Nova Neutron configuration now uses keystone v3 endpoint"
Jenkins [Wed, 24 Feb 2016 10:07:48 +0000 (10:07 +0000)]
Merge "Nova Neutron configuration now uses keystone v3 endpoint"

8 years agoMerge "Update nova::network::neutron variables to drop deprecated parameters"
Jenkins [Wed, 24 Feb 2016 10:04:18 +0000 (10:04 +0000)]
Merge "Update nova::network::neutron variables to drop deprecated parameters"

8 years agoMerge "Allow vncproxy to work with ssl enabled"
Jenkins [Wed, 24 Feb 2016 09:32:46 +0000 (09:32 +0000)]
Merge "Allow vncproxy to work with ssl enabled"

8 years agoAdd UpgradeLevelNovaCompute parameter
Jiri Stransky [Tue, 23 Feb 2016 13:01:59 +0000 (14:01 +0100)]
Add UpgradeLevelNovaCompute parameter

This parameter can be used for pinning (and later unpinning) the Nova
Compute RPC version.

Change-Id: I2f181f3b01f0b8059566d01db0152a12bbbd1c3e

8 years agoIntroduce update/upgrade workflow
Jiri Stransky [Thu, 21 Jan 2016 12:11:23 +0000 (13:11 +0100)]
Introduce update/upgrade workflow

Change-Id: I7226070aa87416e79f25625647f8e3076c9e2c9a

8 years agoAdd resources for major upgrade in Pacemaker scenario
Derek Higgins [Thu, 3 Dec 2015 23:45:15 +0000 (23:45 +0000)]
Add resources for major upgrade in Pacemaker scenario

Add Heat software deployments to be used to upgrade major versions of
OpenStack on the controller nodes. All controller services are taken
down while the upgrade is in progress.

The new updated yum repositories should be configured by another process
e.g. the deployment artifacts transfer via Swift.

Change-Id: Ia0a04e4a11d67e7a5acc53c1f8a8f01ed5ca8675
Co-Authored-By: Giulio Fidente <gfidente@redhat.com>
Co-Authored-By: Jiri Stransky <jistr@redhat.com>
8 years agoNova Neutron configuration now uses keystone v3 endpoint
David Moreau Simard [Thu, 11 Feb 2016 16:55:54 +0000 (11:55 -0500)]
Nova Neutron configuration now uses keystone v3 endpoint

Our current nova-neutron configuration does not work with
the latest puppet-nova. In particular, this patch[1].

This commit adds keystone v3 endpoints to the map and gets the
nova::network::neutron configuration to use them.

[1] https://github.com/openstack/puppet-nova/commit/d09868a59c451932d67c66101b725182d7066a14

Change-Id: Ifb8c23c81c665c2732fa5cd757760668b06a449a

8 years agoAdd meta notify=true to rabbitmq resource
Michele Baldessari [Tue, 23 Feb 2016 07:42:55 +0000 (08:42 +0100)]
Add meta notify=true to rabbitmq resource

See RHBZ 1311005 and 1247303. In short: sometimes when a controller
node gets fenced, rabbitmq is unable to rejoin the cluster. To fix this
we need two steps:
1) The fix for the RA in BZ 1247303
2) Add notify=true to the meta parameters of the rabbitmq resource on
   fresh installs and updates

Note that if this change is applied on systems that do not
have the fix for the rabbitmq resource agent, no action is taken.
So when the resource agent will be updated, the notify
operation will start to work as soon as the first monitor
action will take place.

Fixes RH Bug #1311005

Change-Id: I513daf6d45e1a13d43d3c404cfd6e49d64e51d5a

8 years agoAdd extra config yaml files for big switch agents.
xinwu [Mon, 25 Jan 2016 07:51:37 +0000 (23:51 -0800)]
Add extra config yaml files for big switch agents.

This change adds extra config yaml files for big switch agent
and big switch lldp.

This change is mainly for compute nodes. The changes related
to controller nodes are landed at e78e1c8d9b5a7ebf327987b22091bff3ed42d1c1

This change also removes the neutron_enable_bigswitch_ml2 flag. Instead,
User needs to specify NeutronMechanismDrivers: bsn_ml2 in environment file.

Previous discussion about this change can be found at an abandoned
review request https://review.openstack.org/#/c/271940/

Depends-On: Iefcfe698691234490504b6747ced7bb9147118de

Change-Id: I81341a4b123dc4a8312a9a00f4b663c7cca63d7c

8 years agoUpdate nova::network::neutron variables to drop deprecated parameters
David Moreau Simard [Tue, 9 Feb 2016 22:37:39 +0000 (17:37 -0500)]
Update nova::network::neutron variables to drop deprecated parameters

This commit ensures we are not using any deprecated parameters for
nova::network::neutron and are using the right variable names.

Change-Id: Ic1b41e2cdbb6b180496822cc363c433e9388aa02

8 years agoMerge "Use the class param to configure Cinder 'host' setting"
Jenkins [Fri, 19 Feb 2016 11:29:45 +0000 (11:29 +0000)]
Merge "Use the class param to configure Cinder 'host' setting"

8 years agoMerge "Add TripleO Heat Template Parameters for Neutron Tenant MTU"
Jenkins [Fri, 19 Feb 2016 09:36:07 +0000 (09:36 +0000)]
Merge "Add TripleO Heat Template Parameters for Neutron Tenant MTU"

8 years agoUse the class param to configure Cinder 'host' setting
Giulio Fidente [Thu, 18 Feb 2016 13:34:37 +0000 (14:34 +0100)]
Use the class param to configure Cinder 'host' setting

By configuring the Cinder 'host' setting via the appropriate class
param instead of cinder_config we don't risk to override it if the
user is to pass additional config settings using cinder_config in
ExtraConfig.

Change-Id: Idf33d87e08355b5b4369ccb0001db8d4c3b4c20f

8 years agoMerge "Configure keystone public_endpoint"
Jenkins [Thu, 18 Feb 2016 19:04:48 +0000 (19:04 +0000)]
Merge "Configure keystone public_endpoint"

8 years agoMerge "Enable the ML2 port security extension driver by default"
Jenkins [Thu, 18 Feb 2016 19:03:29 +0000 (19:03 +0000)]
Merge "Enable the ML2 port security extension driver by default"

8 years agoMerge "Add missing : in hieradata key name"
Jenkins [Thu, 18 Feb 2016 16:28:10 +0000 (16:28 +0000)]
Merge "Add missing : in hieradata key name"

8 years agoAdd sysctl settings to disable IPv6 autoconfig and accept_ra
Dan Sneddon [Sun, 7 Feb 2016 15:17:09 +0000 (16:17 +0100)]
Add sysctl settings to disable IPv6 autoconfig and accept_ra

This change adds puppet hieradata settings which disable IPv6
autoconfiguration and accept_ra by default on all interfaces.
When IPv6 is used, the interfaces are individually enabled and
configured with static IP addresses.

The networking on the compute host needs to be completely
separate from the tenant networking, in order to safeguard the
compute host and isolate tenant traffic. This change disables
IPv6 autoconfiguration and acceptance of RAs by default on
interfaces unless specifically enabled.

Without these settings, IPv6 is enabled on all interfaces, as well
as autoconfiguration and accept_ra, so when the compute host
creates a bridge interface for the router (qbr-<ID>), the
compute node will automatically assign an IPv6 address and will
install a default IPv6 route on the bridge interface when it
receives the RAs from the Neutron router.

The change to turn off autoconfiguration means that interfaces
will not self-assign an IPv6 address, and the change to not accept
RAs is a security hardening feature. This requires that a
static gateway address be declared in the network environment
in the parameter ExternalNetworkDefaultRoute. Alternately, sysctl
can be modified to change the accept_ra behavior for specific
interfaces.

Change-Id: I8a8d311a14b41baf6e7e1b8ce26a63abc2eaabef
Closes-bug: 1544296

8 years agoMerge "Make injected CA file readable by others"
Jenkins [Thu, 18 Feb 2016 07:32:55 +0000 (07:32 +0000)]
Merge "Make injected CA file readable by others"

8 years agoMerge "Increase size of connection tracking table"
Jenkins [Thu, 18 Feb 2016 00:45:17 +0000 (00:45 +0000)]
Merge "Increase size of connection tracking table"

8 years agoAdd TripleO Heat Template Parameters for Neutron Tenant MTU
Dan Sneddon [Fri, 29 Jan 2016 02:14:30 +0000 (18:14 -0800)]
Add TripleO Heat Template Parameters for Neutron Tenant MTU

This change adds the TripleO Heat Parameters and Puppet hieradata
to support setting the MTU for Neutron tenant networks. A new
parameter, NeutronTenantMtu is introduced, and this gets used for
the NeutronDnsmasqOptions and in Puppet hieradata.

NeutronTenantMtu is also used in the Puppet hieradata for both the
compute and control nodes. Two values are set:

nova::compute::network_device_mtu

which sets /etc/nova/nova.conf: network_device_mtu = <NeutronTenantMtu>

neutron::network_device_mtu

which sets in /etc/neutron/neutron.conf:
network_device_mtu = <NeutronTenantMtu>

finally, the NeutronDnsmasqOptions parameter becomes a str_format
that maps the NeutronTenantMtu onto the DHCP options,
so a default of 'dhcp-option-force=26,%MTU%' would be formatted to
'dhcp-option-force=26,1300' if NeutronTenantMtu were 1300.

This will set dnsmasq to serve an MTU via DHCP that matches the
NeutronTenantMtu:

/etc/neutron/dnsmasq-neutron.conf:dhcp-option-force=26,1300

Typically, you would change all three of these settings to use small
or jumbo frames in VMs. When using tunneling, NeutronTenantMtu
should be set at least 50 bytes smaller than the physical network
MTU in order to make room for tunneling overhead.

Note that this change does not support setting the MTU on veth
interfaces if veth patches are used to br-int instead of OVS
patches.

Change-Id: I38840e082ee01dc3b6fc78e1dd97f53fa4e63039

8 years agoMerge "Wire the Glance rbd user correctly into the external Ceph template"
Jenkins [Wed, 17 Feb 2016 14:02:39 +0000 (14:02 +0000)]
Merge "Wire the Glance rbd user correctly into the external Ceph template"

8 years agoMake injected CA file readable by others
Juan Antonio Osorio Robles [Wed, 17 Feb 2016 13:48:36 +0000 (15:48 +0200)]
Make injected CA file readable by others

Currently the permissions for the CA file that is injected (if the
environment is set), doesn't permit users that don't belong to the group
that owns the file to read it. This is too restrictive and isn't
necessary, as the certificate should be public.

This is useful in the case where we want a service that can't read the
certificate chain (or bundle) to be able to read that CA certificate.
This is the case for the MariaDB version that is being used in CentOS
7.1 for example.

Change-Id: I6ff59326a5570670c031b448fb0ffd8dfbd8b025

8 years agoMerge "Bind Galera on a hostname for compat with IPv6 addresses"
Jenkins [Wed, 17 Feb 2016 10:11:26 +0000 (10:11 +0000)]
Merge "Bind Galera on a hostname for compat with IPv6 addresses"

8 years agoMerge "Remove start-delay=10s for the Nova resources monitor"
Jenkins [Wed, 17 Feb 2016 10:11:08 +0000 (10:11 +0000)]
Merge "Remove start-delay=10s for the Nova resources monitor"

8 years agoWire the Glance rbd user correctly into the external Ceph template
Giulio Fidente [Tue, 16 Feb 2016 11:41:20 +0000 (12:41 +0100)]
Wire the Glance rbd user correctly into the external Ceph template

We were incorrectly wiring the rbd user to the relevant glance
module parameter, making it was impossible to customize the
rbd user when using an external Ceph.

Change-Id: Ibe4eaedf986a9077f869c6530381e69ee0281f5b

8 years agoMerge "Split pacemaker common check_service function out of _restart.sh"
Jenkins [Tue, 16 Feb 2016 11:23:34 +0000 (11:23 +0000)]
Merge "Split pacemaker common check_service function out of _restart.sh"

8 years agoMerge "Use timeout to check for services status"
Jenkins [Tue, 16 Feb 2016 11:23:26 +0000 (11:23 +0000)]
Merge "Use timeout to check for services status"

8 years agoMerge "Remove DNS hack."
Jenkins [Tue, 16 Feb 2016 11:13:40 +0000 (11:13 +0000)]
Merge "Remove DNS hack."

8 years agoMerge "Update Dell Storage Center api port setting"
Jenkins [Mon, 15 Feb 2016 22:47:19 +0000 (22:47 +0000)]
Merge "Update Dell Storage Center api port setting"

8 years agoMerge "Switch to POLL_TEMP_URL for config transport"
Jenkins [Mon, 15 Feb 2016 21:53:05 +0000 (21:53 +0000)]
Merge "Switch to POLL_TEMP_URL for config transport"

8 years agoMerge "Minor fixes to allow local docker registry usage"
Jenkins [Mon, 15 Feb 2016 19:24:01 +0000 (19:24 +0000)]
Merge "Minor fixes to allow local docker registry usage"

8 years agoMerge "Enable SSL middleware for cinder"
Jenkins [Mon, 15 Feb 2016 15:16:20 +0000 (15:16 +0000)]
Merge "Enable SSL middleware for cinder"

8 years agoUpdate the path to the Docker file
Dougal Matthews [Wed, 6 Jan 2016 11:29:47 +0000 (11:29 +0000)]
Update the path to the Docker file

The file was renamed in db16fd6b59257ea9eacbf071e9e799041822dcab
(Change-Id I7837ed7ed3e807ec5c1276904893695918bef293).

Change-Id: Ia8bdd705fddf00acc20116d21f39cb80b9fd693e

8 years agoMerge "Update the capabilities map file name to be more consistent"
Jenkins [Fri, 12 Feb 2016 21:29:09 +0000 (21:29 +0000)]
Merge "Update the capabilities map file name to be more consistent"

8 years agoMinor fixes to allow local docker registry usage
Jeff Peeler [Fri, 8 Jan 2016 17:20:53 +0000 (12:20 -0500)]
Minor fixes to allow local docker registry usage

Changed the heat-docker-agents namespace to use the namespacing
specified in the environment file, which reduces modifications required
on the user when using a local registry.

Changed the start agents script to handle using a local registry both
with a namespace and without.

Change-Id: I16cc96b7ecddeeda07de45f50ffc6a880dabbba6

8 years agoAdd missing : in hieradata key name
James Slagle [Fri, 12 Feb 2016 18:08:37 +0000 (13:08 -0500)]
Add missing : in hieradata key name

This hieradata key, neutron::agents::ml2::ovs:bridge_mappings was
missing a : before bridge_mappings causing the value to be blank in
/etc/neutron/plugins/ml2/openvswitch_agent.ini even if a value had been
specified.

Change-Id: I377565d3fb821be1bb2dc7d92ec1ad25a4a3b1f1

8 years agoRemove DNS hack.
Ian Main [Fri, 12 Feb 2016 15:11:08 +0000 (10:11 -0500)]
Remove DNS hack.

With a properly configured undercloud the DNS is fine.  We can remove
the 8.8.8.8 dns setting.

Change-Id: I8ba98e76f95fd0a6f3f34cb5578e6c3ea7a1d15e

8 years agoMerge "Nova now requires an api database to be created"
Jenkins [Fri, 12 Feb 2016 13:19:44 +0000 (13:19 +0000)]
Merge "Nova now requires an api database to be created"

8 years agoRemove start-delay=10s for the Nova resources monitor
Giulio Fidente [Fri, 22 Jan 2016 14:16:05 +0000 (15:16 +0100)]
Remove start-delay=10s for the Nova resources monitor

As per conversation in [1], these settings should have probably never
been there.

1. https://bugzilla.redhat.com/show_bug.cgi?id=1262409

Change-Id: I116f825ba0fe3e4faac8dd347bb087e1b4c70e57

8 years agoEnable heat-manage purge_deleted cron job
Steve Baker [Fri, 12 Feb 2016 01:07:25 +0000 (14:07 +1300)]
Enable heat-manage purge_deleted cron job

Without this the heat database tables will grow without limit.

Change-Id: I687e733db1a73ebc2047609a03be768093010dd4
DependsOn: Ia2b80e5003450cd794ebb0c9ca72200ec8616e81

8 years agoMerge "Increase default Cinder LVM backing file to 10G"
Jenkins [Thu, 11 Feb 2016 21:29:13 +0000 (21:29 +0000)]
Merge "Increase default Cinder LVM backing file to 10G"

8 years agoMerge "puppet: run keystone in wsgi"
Jenkins [Thu, 11 Feb 2016 20:22:31 +0000 (20:22 +0000)]
Merge "puppet: run keystone in wsgi"