| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
This is a first implementation of Ceph support in TripleO with Puppet:
* Install ceph-mon on controller node
* Install ceph-osd on cephstorage node
Co-Authored-By: Giulio Fidente <gfidente@redhat.com>
Change-Id: I48488cbe950047fae5e746e458106d6edb9a6183
|
|
|
|
|
|
|
|
|
|
|
|
This patch moves all the related mergepy templates for the
overcloud into a deprecated directory. The Makefile has
been updated so that overcloud.yaml is still generated
at the top level so this shouldn't break end users.
This is to reduce confusion for new users who are learning the
TripleO heat templates and find the fact that we have two full
implementations very confusing.
Change-Id: I0848aca4dee3e37cb4c6089c5f655ad22ac6c5fd
|
|
This patch applies the allNodesConfig data to swift storage
nodes. This contains hosts information which could be useful.
Change-Id: Iaccfdc698e371d6618d561c33f256ccc3c166fb7
|
|
This patch adds a new BlockStoreNodesPostDeployment resource
which can be used along with the environment file to
specify a nested stack which is guaranteed to execute
after all the BlockStore config deployments have executed.
This is really useful for Puppet in that Heat actually
controls where puppet executes in the deployment
process and we want to ensure puppet runs after
all hiera configuration data has be deployed to
the nodes. With the previous approach some of the
data would be there, but allNodes data would not be
guaranteed to be there in time.
As os-apply-config (tripleo-image-elements) have their
ordering controlled within the elements themselves an empty stubbed
in nested stack has been added so that we don't break that
implementation.
Change-Id: I29b3574e341eecd53b2867788f415bff153cfa9f
|
|
This patch adds a new ObjectStoreNodesPostDeployment resource
which can be used along with the environment file to
specify a nested stack which is guaranteed to execute
after all the ObjectStore config deployments have executed.
This is really useful for Puppet in that Heat actually
controls where puppet executes in the deployment
process and we want to ensure puppet runs after
all hiera configuration data has be deployed to
the nodes. With the previous approach some of the
data would be there, but allNodes data would not be
guaranteed to be there in time.
As os-apply-config (tripleo-image-elements) have their
ordering controlled within the elements themselves an empty stubbed
in nested stack has been added so that we don't break that
implementation.
Change-Id: I778b87a17d5e6824233fdf9957c76549c36b3f78
|
|
This patch adds a new ComputeNodesPostDeployment resource
which can be used along with the environment file to
specify a nested stack which is guaranteed to execute
after all the Compute config deployments have executed.
This is really useful for Puppet in that Heat actually
controls where puppet executes in the deployment
process and we want to ensure puppet runs after
all hiera configuration data has be deployed to
the nodes. With the previous approach some of the
data would be there, but allNodes data would not be
guaranteed to be there in time.
As os-apply-config (tripleo-image-elements) have their
ordering controlled within the elements themselves an empty stubbed
in nested stack has been added so that we don't break that
implementation.
Change-Id: I80bccd692e45393f8250607073d1fe7beb0d7396
|
|
This patch splits out the BootstrapNode config
such that alternate implementation (puppet for example)
can implement their own SoftwareConfig's via a nested stack.
This is controlled by the standard overcloud heat environment.
For os-apply-config deployments the implementation should work the
same as before.
For puppet deployments the implementation uses hiera metadata
to configure bootstrap_nodeid.
Change-Id: I691a9d7c474866038a5d47beab295899b5479d03
|
|
Allow to install & configure RabbitMQ in cluster with Puppet on
the controller node.
Change-Id: Iebbf55c75b8c80453c7313bb41faf42c7fdf7159
|
|
This patch splits out the allNodesConfig config
such that alternate implementation (puppet for example)
can implement their own SoftwareConfig's via a nested stack.
This is controlled by the standard overcloud heat environment.
For os-apply-config deployments the implementation should work the
same as before.
For puppet deployments the implementation uses hiera metadata
to configure rabbit_nodes. The puppet deployment doesn't support
hosts, or freeform sysctl metadata yet so those are the same
for now as well.
Change-Id: I34ae30b1f37aca8b39586f7e350511462d66f694
|
|
This reverts commit 4d470abc589c660cd55e4ced92de234fdf83d882
where we disabled swift (and the glance swift backend) due
to the fact that some of the Heat metadata wasn't showing up.
Change-Id: Ib0c01be5844aa79d74b7de02ba3d0657db5047ba
Closes-bug: 1418805
|
|
This patch splits out the SwiftDevicesAndProxy config
such that alternate implementation (puppet for example)
can implement their own SoftwareConfig's via a nested stack.
This is controlled by the standard overcloud heat environment.
For os-apply-config deployments the implementation should work the
same as before.
For puppet deployments the implementation uses hiera metadata
to configure swift devices.
Partial-bug: 1418805
Change-Id: Ibf6038460f36279ad51a04947589d4a03a553f66
|
|
This patch adds a new ControllerNodesPostDeployment resource
which can be used along with the environment file to
specify a nested stack which is guaranteed to execute
after all the Controller config (HA, or other) have
executed.
This is really useful for Puppet in that Heat actually
controls where puppet executes in the deployment
process and we want to ensure puppet runs after
all hiera configuration data has be deployed to
the nodes. With the previous approach some of the
data would be there, but most of the HA data which
actually gets composed outside of the controller-puppet.yaml
nested stack would not be guaranteed to be there in time.
As os-apply-config (tripleo-image-elements) have their
ordering controlled within the elements themselves an empty stubbed
in nested stack has been added so that we don't break that
implementation.
Partial-bug: 1418805
Change-Id: Icd6b2c9c1f9b057c28649ee3bdce0039f3fd8422
|
|
This cleans up the top level tree by moving all the puppet
related bits into the puppet directory. The only exception
is overcloud-resource-registry-puppet.yaml which is
the puppet environment file and is used externally.
Change-Id: Idb65a7143b0f29e5579d4e9d1642e4cda6f65d50
|
|
The new ceph-source.yaml file provides the config settings needed
by the elements which configure Ceph on controllers (monitors) and
storage nodes (OSDs) as well as the Cinder backend which uses it.
There is also a without-mergepy copy named ceph-storage.yaml
Change-Id: I954861536c41b2a7e6cbd86a0f0b55004eed4c70
|
|
Not all installations have an NtpServer configured and if
they don't the ntp service will fail to startup correctly.
This patch makes it so that ntp is only enabled if
the ntp::servers array is greater than 0.
Change-Id: I8417f87ad2a3c1237ebb00ee1232b5313cd45d46
|
|
We have an issue where swift.devices metadata isn't showing
up on our controllers. This causes ringbuilding to fail
meaning swift-proxy won't startup.
This patch disables the swift-proxy and glance swift backend
until we can figure out exactly what caused this change.
Change-Id: I723a4b703d979d7475ac48f41c4c0ac91c306884
Partial-bug: 1418805
|
|
This adds an option which enables package installation via
Yum when Puppet executes. Users might want to disable Yum
installation of packages via puppet when using pre-installed
images.
The option is off by default: meaning that Puppet will no
longer install packages by default. Users will need to
enable the EnablePackageInstall in order to get
the previous behavior.
The intent is to use the default_parameters section
of the Heat environment to allow users to cleanly enable this
features without wiring it into the top level. This is because
the new parameter is Puppet specific and doesn't really apply to
other implementations. Kilo Heat already has support for
default_parameters and so does python-heatclient.
NOTE: most TripleO users do not yet have the heatclient
features because setup-clienttools in tripleo-incubator only installs
releases via pip. It is for these reasons the default_parameters
section in overcloud-resource-registry-puppet.yaml is commented out
for now.
Change-Id: I3af71b801b87d080b367d9e4a1fb44c1bfea6e87
|
|
This configures an snmp agent for the undercloud
ceilometer 'hardware' metering. This rely's on the
razorsedge/puppet-snmp which we are adding in
I8ae104de7382767c3448a493cd37ff2994cf4f52.
Change-Id: If2b6b63279b9b0402c5136ff1635e10acad1de7e
|
|
This patch updates puppet on the controller so that it
configures the Neutron dnsmasq options file data with
the value provided by the Heat NeutronDnsmasqOptions
parameter.
Properly configuring this setting can help resolve/tune
overcloud instance connectivity issues w/ SSH etc.
Change-Id: If47ab3d3002ebe19fc980ca5d37f84f4d8851f9b
|
|
This updates the controller config for the Neutron metadata
agent so that it configures the correct nova_metadata_ip
in /etc/neutron/metadata_agent.ini.
Change-Id: I4d6658ba54e582673938fa14a8c7de287dcd6662
Closes-bug: #1416781
|
|
This moves the loadbalancer composition class parameters
into the loadbalancer specific software deployment.
This keeps the resource contained and seperate from the rest
of the OS hiera data configs.
Change-Id: I8af48b479348e431a8e563917e1345ca4b895a60
|
|
This patch adds the ability to configure the Heat API and
Heat engine on controller nodes via puppet.
Change-Id: Ie81090bceed3e18199a36ebb11d1cbcaea83c410
|
|
This patch adds NTP support to all roles.
As part of this change overcloud-without-mergepy.yaml has
also been updated so that it passes the NtpServer parameters into
the Swift and Cinder storage node templates so that Ntp can
also be configured on those machines as well.
NOTE: The puppet support here uses the puppetlabs-ntp modules
which we add in Ib10ccbfdb3140b19f40049707548c6655d250e1c.
Change-Id: If2ef236fa42a714e84c6944eee5fe4daddf3fedf
|
|
This patch adds support for the Ceilometer controller
role including the Ceilometer:
-API
-central agent
-alarm notifier
-alarm evaluator
-collector
-expirer
In order to enable swift metering the swift::proxy ceilometer middleware
was added in.
Also, a minor adjustment to the existing ceilometer HA proxy setting
was made to accommodate ceilometer auth settings. (not exactly sure
why but this seems to be required)
Like upstream TripleO Ceilometer is currently using a MySQL database
backend. A follow on patch can support configuring MongoDB for use
with Ceilometer.
Change-Id: I4e171274bd7679d386d93492d13dfa7c5d37f6a8
|
|
Cinder block storage nodes shouldn't need to know the
AdminPassword and CinderPassword values. There are
no services which require Keystone related passwords
on the block storage nodes.
Change-Id: I4aa89347c60ec6258bd66725a895f6fd2b4844f6
|
|
This patch implements the required changes to configure
common Cinder block storage nodes via Puppet.
Change-Id: Iac8b4679a00f58d5faac4a1d08b7a830f0360ba5
|
|
Our existing default (replicas == 1) means that no data
(or copies) is being replicated in a multi-node Swift
environment. This seems like a bad production default
setting and could easily slip by if not set.
Setting it to 3 shouldn't hurt anything and seems to follow
suit with what several production installers (based around Puppet)
actually use. If using an installation with less than 3 swift
nodes I believe swift will do its best, and still work fine.
FWIW I noticed this when testing a multi-node Puppet swift
installation and was surprised when I didn't see any *data
files getting replicated across the storage cluster.
Change-Id: I44bdfff7aae6bdf845b79ca1f8f450c22113caed
|
|
In doing the Puppet version of the Swift role I noticed
4 parameters which we apply to storage nodes which should
not be required. This patch drops the following parameters
from the swift-storage and swift-storage-puppet nested
stacks which should not be required.
1) ControllerIP: There is no reason a storage node should need
the IP address of the controller. The swift proxy would need
this information in order to be able to contact keystone.
This swift-proxy is not installed on storage nodes so we can
drop the parameter here.
2) NeutronEnableTunnelling: There is no reason for Neutron
to be installed on Swift storage nodes. No need to create
an OVS bridge either.
3) NeutronNetworkType: Similar to above. No neutron requirements
exist here so this parameter is not required.
4) Password: This only applies to the the swift-proxy which is
currently part of our controller role. Storage nodes shouldn't need
the keystone service-password for any reason.
Change-Id: Icbf05363475c388fc722277da3d3d00a7355b19a
|
|
Now that we have swift we can switch glance over
to make use of it.
Change-Id: I9513cb63079235337b684aa734af73a0f0cc0afd
|
|
This patch implements the required changes to configure
swift storage nodes via Puppet. Similar to the overcloud
we generate the rings on each node (with the same seed).
Change-Id: I677c85b09b6e656b3ac1f938a4bd6bc7daae1755
|
|
This patch adds support for a Swift proxy and storage
node on the controller.
The implementation is fairly straightforward with the
exception of building the ring. I've followed an
upstream TripleO model here where we build the
actual ring on each node (rather than build once
and rsync). This works because Heat will always
know all the devices ahead of time. In the future
when we have Heat breakpoints it might be possible
to consider optimizing this by generating the ring
once and then rsyncing to all the nodes.
The ringbuilder logic is executed as a seperate
Heat software deployment. On the controller the ring
is executed in between the base service (mysql/rabbit)
and OpenStack service steps. This is to ensure the
ring exists before the Swift proxy is started.
Having the ringbuilder.pp logic as a separate software
config should allow us to reuse it for the Storage
node role.
It should also be noted that swift.zones support is
added here but we are missing an upstream Heat
template change in order for it to be wired
in properly. See: I0e0f5189da1575f2e1ed7fba4bbbe13a8fbf6221
Likewise we need to properly wire in SwiftRingBuild as well.
See: I01311ec3ca265b151f8740bf7dc57cdf0cf0df6f
The underlying puppet ringbuilder code is already wired
to support this change when it lands.
As is this works today and will provide a working Overcloud
Swift-proxy/storage node config. Will follow this up with
a related Swift storage node patch which should allow
puppet to be used for configuration on the storage nodes
as well...
Change-Id: Id1272f796e2507a7357309e8cd6a51ad9e0160af
|
|
In I250dc1a8c02626cf7d1a5d2ce92706504ec0c7de we split
out just the Controller software config in an effort
to provide hooks for alternate implementations (puppet).
This sort of worked but caused quirky ordering issues
with signal handling. It also causes problems for Tuskar
which would prefer to think of these nested stacks and
not have us split out just the software configs like this.
This patch moves all the compute related stuff for
our two implementations:
compute.yaml: is used by os-apply-config (uses the
tripleo-image-elements)
compute-puppet.yaml: uses stackforge puppet-* modules for
configuration
By duplicating the entire compute in this manner we make
it much easier to create dependencies and implement proper
signal handling. The only (temporary) downside is the duplication
of parameters most of which will eventually go away when we move
using the global parameters via Heat environment files instead.
Change-Id: I49175d1843520abc80fefe9528442e5dda151f5d
|
|
In I228216a0b55ff2d384b281d9ad2a61b93d58dab9 we split
out just the Controller software config in an effort
to provide hooks for alternate implementations (puppet).
This sort of worked but caused quirky ordering issues
with signal handling. It also causes problems for Tuskar
which would prefer to think of these nested stacks and
not have us split out just the software configs like this.
This patch moves all the controller related stuff for
our two implementations:
controller.yaml: is used by os-apply-config (uses the
tripleo-image-elements)
controller-puppet.yaml: uses stackforge puppet-* modules for
configuration
By duplicating the entire controller in this manner we make
it much easier to create dependencies and implement proper
signal handling. The only (temporary) downside is the duplication
of parameters most of which will eventually go away when we move towards
using the global parameters via Heat environment files instead.
Change-Id: Iaf3c889d7c8815f862308cd8e15ce1010059f5c6
|
|
|
|
|
|
|
|
|
|
This change will allow for the enablement of Neutron routers HA
via the new NeutronL3HA parameter.
Change-Id: Ia5f7c0b4e89159456482e840c50d166ec5f25d4c
Implements: blueprint tripleo-icehouse-ha-production-configuration
|
|
This was added in I36fece56bafa9fe9c4883b572687b3fc819eeae1
and is missing from overcloud-without-mergepy.
Change-Id: I5c2566cc77247574f8d687eaab8094de481a8c67
|
|
This was added in Icc5e431a7e2884b3ca3a255b6fd901619bc98460
and is missing from overcloud-without-mergepy.
Change-Id: I1273b646c04783712fd3f8baccafead11817689c
|
|
|
|
We have never created these additional storage nodes by default with
the old templates; we agreed on adding a job for this in CI [1] so
we will override the default value in the specific CI job.
1. https://github.com/openstack-infra/tripleo-ci/blob/master/docs/wanted_ci_jobs.csv
Change-Id: Iaec38807bc209fc28d83e3d6922269e803110053
|
|
Currently the all templates have an invalid setting for NTP
setup for the fudge setting. This should be removed from
the templates which will remove the warning seen in syslog.
ntpd[...]: inappropriate address xxx.xxx.xxx.xxx for the
fudge command, line ignored
Partial-Bug: 1408379
Relates-To: Ib9931b84925d9ceb32f18e9adc5be64402fbf61e
Change-Id: I56a03dc0a899a8c515f2a05d678d7e80e9b7b93c
|
|
This patch provides an alternate implementation of
the OS::TripleO::Controller::SoftwareConfig which uses Puppet
to drive the configuration. Using this it is possible
to create a fully functional overcloud controller instance
which has the controller node configured via Puppet
stackforge modules. Initially this includes only the
following services:
MySQL
RabbitMQ
Keepalived/HAProxy (HA is not yet fully supported however)
Nova
Neutron
Keystone
Glance (file backend)
Cinder
Using these services it is possible to run devtest_overcloud.sh
to completion. The idea is that we can quickly add more
services once we have CI in place.
In order to test this you'll want to build your images
with these elements:
os-net-config
heat-config-puppet
puppet-modules
hiera
None of the OpenStack specific TripleO elements
should be used with this approach (the nova/neutron
elements were NOT used to build the controller image).
Also, rather than use neutron-openvswitch-agent to configure
low level networking it is recommended that os-net-config
by configured directly via heat modeling rather than
parameter passing to init-neutron-ovs. This allows us to
configure the physical network while avoiding the coupling to
the neutron-openvswitch-element that our standard
parameter driven networking currently uses. (We still need
to move init-neutron-ovs so that it isn't coupled and/or deprecate
its use entirely because the heat drive stuff is more flexible.)
Packages may optionally be pre-installed via DIB using the
-p option (-p openstack-neutron,openstack-nova) etc.
Change-Id: If8462e4eacb08eced61a8b03fd7c3c4257e0b5b8
|
|
|
Emilien Macchi
Jenkins
Dan Prince
Giulio Fidente
Nicholas Randon