1) Update doc links according to OpenStack document migration
2) Use https instead of http for docs links
Change-Id: Id87e91d1f95bcb94c31125c467abab17242cb36f
Mark OpenShift templates as deprecated as they have not been tested on
an ongoing basis and are now superseded.
Change-Id: Ia8c4de847787ff3fa624a175f5d9d0305ac420c9
This change corrects many problems detected by the `yamllint` linter.
It's a preparation for enabling this linter in change
Ie746230f28fe3ed0cf218201d5a3810f7bc44070.
For instance, the first run of the YAML linter helped discovering a key
duplication problem in `oso_ha.yaml`: the `depends_on` key was present
twice, so the first occurence was ignored.
Other changes are cosmetic: extra spaces, extra blank lines, missing
newlines at end-of-file, etc.
Change-Id: I7f2369adfb152fd2a74b9b105e969e653e592922
In mitaka, a new feature is introduced to ignore a given
set of errors and is used here to fix the zuul faliure.
In addition, it fixes the invalid template errors.
For those templates which causes circular error is
moved to invalid folder
NOTE: openshift-origin/centos65/highly-available is marked
as invald to make the build jobs to pass.
depends-on: I4a3043fd17b69a346d447dfbd17488040cf9b387
Change-Id: I7b5b72cb8f6ff53b82edf92799a57917c718e032
Closes-bug: #1554380
We need the latest version of puppet-release or we would not be
able to pull up the right ruby version to install OpenShift.
Change-Id: Ic7ad9ff7645c97133f4f69df839c9a2d33098476
Since autoscaling group name contributes to hostname, and there should
have no underscore in it so we remove the underscore.
Change-Id: I5e5848ed5f132f06ebf601f73369653c807e413d
Partial-Bug: #1394096
New heat templates for Openshift-Origin deployment using scalability functionality.
Templates are based on the Openshift-Origin template
(https://github.com/openstack/heat-templates/tree/master/openshift-origin/F19).
The aim is to add scalability functionality to this template. Metric use
for the scalability are cpu_util and nbgears. This second metric is
a custom metric implemented by a bash script calling by cron to get
from the broker node the number of gears used and the number of node
connected, to determine when a new node is needed. The data is send to
the Ceilometer V2 API and used in the heat template.
To avoid compatibility issues due to changes in this project or
in Puppet modules it depends, versions parameters are provided in hot templates.
I implemented 5 templates :
aws format :
- openshift (1 node and 1 broker)
hot format :
- openshift (1 node and 1 broker)
- openshift-scalable-cpu (scalable openshift environment using cpu_util metric)
- openshift-scalable-nbgears (scalable openshift environment using nbgears custom metric)
- openshift-scalable-cpu-nbgears (scalable openshift environment using cpu_util metric and nbgears custom metric)
Change-Id: Idfdf158ff6176c5ef6902e27a6e764e23171cf4c
Implements: blueprint heat-openshift-templates
Fix issues introduced recently in template validation: a typo in a
wordpress template and environments files not named properly.
Change-Id: I8c3c11cfac841c02e700c62b8365821519985267
These templates are heavily based on the existing OpenShift enterprise
templates for use with RHEL. The OpenShift installer utilized in these
templates has been slightly modified to support CentOS as well as make
the install completely automated (aside from creating districts if so
desired.)
Change-Id: I74acfdd553eb6a4c7ac771b6c0ec6543e1e63ea9
The puppet-openshift_origin module recently went through a
refactor in preperation for the v4 OSO release. This patch
changes the necessary puppet parameters for the template to
continue functioning.
Change-Id: Ia9145ebf744e45ee3c7341a04f05d257515a976d
The yum repo files had insecure settings for sslverify and
gpgcheck. Also change references to http to https.
Change-Id: I10cbc33a734f9ed36ef6f22174c65efaf3b659d7
This adds a new template for deploying OpenShift Origin on CentOS. In
this case two hosts are used with separated node and broker roles.
Change-Id: Iabf125b2ea9c0e0d45dfb9444017de6d55798d1f
Constraints was misspelled, however the default value would work fine if
instance_type wasn't specified. The list syntax for HOT was also
incorrect.
There was a dependency problem that is fixed now, so force installing
nodejs-bson is no longer required.
Change-Id: Iabc9ff4be7cfa479d6cf11f735857f16b3de4905
Previously, the Node Wait Handle/Condition referenced the Broker
instead of the Node. This caused the Heat stack to show complete
prior to the Node being truly configured and ready.
Change-Id: I7ab1dce8de0b892e3495ef16c26dc8e8bd0475b7
Previously, neutron security rules were incorrectly specified
to allow ActiveMQ TCP ports. This was not previously observed
since a single OpenStack Compute Host was used to run Broker and
Node instances. Additionally, duplicate entries for ActiveMQ
TCP port 8161 were specified. This patch removes the duplicate
entries and changes the TCP port number from ActiveMQ to
Mcollective (61613).
Change-Id: Ia35267a4485bba4aab36b9a629993ab94f42ad11
Previously, neutron security group tcp/udp port numbers were
defined as strings. This was not an issue when testing with trunk
but is an issue when testing with 2014.1.b2.
2014.1.b2 expects the parameters to be defined as integers not
strings. When the tcp/udp ports are defined as integers, the
security group rules work with trunk as well as 2014.1.b2.
Change-Id: If7e5af163993b4cad3f31583ea73e9e856ab0ccb
Previously, the OSO broker/node instances would get spawned
using a combination of stack name, nova instance name and UUID.
This patch allows a user to specify a hostname for broker and
node instances. It makes working with broker/node instances
much easier from a user's perspective.
Change-Id: Ib077acb257e5add6c36b4a913cc6752c716524a7
This has been tested on CentOS 6.5 as far as the deployment guide
indicates in section 11:
http://openshift.github.io/documentation/
oo_deployment_guide_comprehensive.html#testing-the-configuration
Change-Id: I40766a46fb45e618e35a4f7a3662bd693ac8613c
heat-cfntools from the tripleo-image-elements repo is a bit heavyweight.
Instead this element simply installs the latest repo package of heat-cfntools
and symlinks to /opt/aws/bin.
Change-Id: I5a0ec433c352f3ac0bb415381c7e41d4370d1e85
F18 builds are unreliable (or for some people don't work at all),
this seems due to the following bug:
https://bugzilla.redhat.com/show_bug.cgi?id=906031
The simplest workaround is to use the netinst ISO instead of the DVD
so we pick up the newer curl package which fixes the problem
Fixes bug #1185810
Change-Id: I066356e8a780caa8ca2798644ec7594b0536005c
Broker has scripts /etc/heat/notify-scale-up and /etc/heat/notify-scale-down
which will trigger heat autoscaling.
Scaling decisions will be made within the broker itself using its own
platform agnostic scaling scripts:
https://github.com/openshift/openshift-extras/tree/enterprise-1.1/node-manager
This template has successfully scaled up from 1 to 3 nodes, and scaled back
down to 1 by manually executing notify-scale-up and notify-scale-down.
Change-Id: Idb9920438bbe598b610efc8eca5bc158181b74a7
This deploys an identical stack to the original json.
OpenShiftAutoScaling.template has been deleted to reduce
maintenance burden.
Change-Id: Ie2840f65ed0d6f6964b4be71d6b19368e844bac4
Updates OpenShiftAutoScaling.template with fixes contained in
OpenShift.template.
Nodes do scale when the hard-coded cfn-push-stats value is raised
above the threshold.
Change-Id: I79051a15a5116034d06fd95b5cc8c3412eda33a8
These have been used to deploy a working openshift stack.
They pre-install more images than the oz templates, based on the yum
log of extra installs during boot time.
Change-Id: Icb246033964e1388aeebdd4229790acfe0e17718
With these changes it is now possible to launch the wordpress
gear on the node.
Changes include:
- set environment variables at the start of the UserData and use them
throughout instead of heat refs
- set /etc/resolv.conf to point at the broker's bind
- on the broker run setsebool -P httpd_unified=on
oo-accept-broker suggested this was necessary, and the console errored
until it was set
- enforce selinux, it was left in permissive mode
- setsebool httpd_unified needs to be set on the broker
- node /etc/resolv.conf needs to be overwritten after puppet apply
- Outputs for console URL and nameserver entry for DNS resolution of apps
Change-Id: I28c9ce4b19f4a0aa59ef514a269ae30b6fb9f6fa
This includes the following changes:
- requests from outside the broker were not reaching it without
setting configure_firewall => true
- puppet module install was failing due to missing /etc/puppet/modules
- on the node puppet --apply was failing because the openshift_origin
module wasn't installed
Change-Id: Ie6f1f0f6e2bdbde1521ffdd285041c87f3933e42
Remove explicit version dependency for old boto version, as this
won't work with the Fedora packaged cloud-init, and heat no longer
requires this version, so we should use the Fedora boto package
Change-Id: Iaf340b4793fdf7ffc3f9b3d0197ca9f6ba6f42fe