apex-puppet-tripleo.git
7 years agoHAProxy: Set listen options for internal services too
Juan Antonio Osorio Robles [Tue, 8 Aug 2017 14:40:29 +0000 (17:40 +0300)]
HAProxy: Set listen options for internal services too

This was missed from a previous commit, as described in the bug report.
We need to set this variable in this case as well, else it will use the
undefined variable, thus ignoring anything that the user had set.

Change-Id: I6810e7bb3eed16a6478974ac759c3f720a41120a
Closes-Bug: #1709332

7 years agoMerge "Enable TLS configuration for containerized RabbitMQ"
Jenkins [Thu, 10 Aug 2017 15:26:17 +0000 (15:26 +0000)]
Merge "Enable TLS configuration for containerized RabbitMQ"

7 years agoMerge "Use clustercheck credentials to poll galera state in container"
Jenkins [Wed, 9 Aug 2017 23:13:52 +0000 (23:13 +0000)]
Merge "Use clustercheck credentials to poll galera state in container"

7 years agoMerge "Enable innodb_buffer_pool_size configuration"
Jenkins [Wed, 9 Aug 2017 13:51:16 +0000 (13:51 +0000)]
Merge "Enable innodb_buffer_pool_size configuration"

7 years agoEnable TLS configuration for containerized RabbitMQ
Damien Ciabrini [Mon, 7 Aug 2017 20:32:51 +0000 (20:32 +0000)]
Enable TLS configuration for containerized RabbitMQ

In non-containerized deployments, RabbitMQ can be configured to use TLS for
serving and mirroring traffic.

Fix the creation of the rabbitmq bundle resource to enable TLS when configured.
The key and cert are passed as other configuration files and must be copied by
Kolla at container startup.

Change-Id: Ia64d79462de7012e5bceebf0ffe478a1cccdd6c9
Partial-Bug: #1709558

7 years agoMerge "Update swift-proxy unit tests for puppet5"
Jenkins [Mon, 7 Aug 2017 16:13:48 +0000 (16:13 +0000)]
Merge "Update swift-proxy unit tests for puppet5"

7 years agoMerge "Update link addresses in README.md"
Jenkins [Mon, 7 Aug 2017 03:38:43 +0000 (03:38 +0000)]
Merge "Update link addresses in README.md"

7 years agoUpdate swift-proxy unit tests for puppet5
Emilien Macchi [Mon, 7 Aug 2017 03:33:40 +0000 (20:33 -0700)]
Update swift-proxy unit tests for puppet5

The latest version of puppet now require the class dependencies included
in the unit tests.

Change-Id: I0b6462f697f2d8012f8a785660c004f3efb13fdc

7 years agoMerge "Enable encryption of pacemaker traffic by default"
Jenkins [Sat, 5 Aug 2017 07:33:55 +0000 (07:33 +0000)]
Merge "Enable encryption of pacemaker traffic by default"

7 years agoMerge "Update openstackdocstheme>=1.16.0"
Jenkins [Sat, 5 Aug 2017 07:19:03 +0000 (07:19 +0000)]
Merge "Update openstackdocstheme>=1.16.0"

7 years agoMerge "Configure dockerd with --iptables=false"
Jenkins [Fri, 4 Aug 2017 16:36:45 +0000 (16:36 +0000)]
Merge "Configure dockerd with --iptables=false"

7 years agoMerge "Ensure directory exists for certificates for haproxy"
Jenkins [Fri, 4 Aug 2017 11:04:00 +0000 (11:04 +0000)]
Merge "Ensure directory exists for certificates for haproxy"

7 years agoEnable innodb_buffer_pool_size configuration
Mike Bayer [Wed, 2 Aug 2017 14:38:45 +0000 (10:38 -0400)]
Enable innodb_buffer_pool_size configuration

Adds a hiera-enabled setting for mysql.pp to
allow configuration of innodb_buffer_pool_size, a key
configurational element for MySQL performance tuning.

Change-Id: Iabdcb6f76510becb98cba35c95db550ffce44ff3
Closes-bug: #1704978

7 years agoConfigure dockerd with --iptables=false
Dan Prince [Wed, 2 Aug 2017 20:51:47 +0000 (16:51 -0400)]
Configure dockerd with --iptables=false

This change defaults --iptables=false for dockerd to avoid
having Docker create its own FORWARD iptables rules. These
rules can interact with normal OS networking rules and disable
communications between hosts on reboot.

Change-Id: I875fa14f7d810c7f0aba3b3a1b04b60a19470f0f
Closes-bug: #1708279

7 years agoUpdate link addresses in README.md
zhangdebo1987 [Thu, 3 Aug 2017 06:21:14 +0000 (14:21 +0800)]
Update link addresses in README.md

Change-Id: I0ad611bd669e9fb5f119237034dca347641c74b5

7 years agoUse normal socket file permissions instead of polkit
Oliver Walsh [Tue, 25 Jul 2017 20:05:35 +0000 (21:05 +0100)]
Use normal socket file permissions instead of polkit

The default (on RHEL/CentOS) is to use polkit but this is only useful
for GUI support or for fine grained API access control.  As we don't
require either we can achieve identical control using plain old unix
filesystem permissions.

I've merged Sven's changes from https://review.openstack.org/484979
and https://review.openstack.org/487150.

As we need to be careful with the libvirtd option quoting I think it's
best to do this in puppet-tripleo instead of t-h-t yaml.

The option to override the settings from t-h-t remains.

Co-Authored-By: Sven Anderson <sven@redhat.com>
Reverts I91be1f1eacf8eed9017bbfef393ee2d66771e8d6

Closes-bug: 1696504

Change-Id: I507bdd8e3a461091562177403a2a55fcaf6694d2
Depends-On: I17f6c9b5a6e2120a53bae296042ece492210597a

7 years agoMerge "Support for Dell EMC Unity Cinder Driver"
Jenkins [Wed, 2 Aug 2017 08:46:27 +0000 (08:46 +0000)]
Merge "Support for Dell EMC Unity Cinder Driver"

7 years agoEnsure directory exists for certificates for haproxy
Juan Antonio Osorio Robles [Wed, 2 Aug 2017 07:19:48 +0000 (10:19 +0300)]
Ensure directory exists for certificates for haproxy

We used to rely on a standard directory for the certificates and keys
that are requested by certmonger. However, given the approach we plan to
take for containers that's described in the blueprint, we need to use
service-specific directories for the certs/keys, since we plan to
bind-mount these into the containers, and we don't want to bind mount
any keys/certs from other services.

Thus, we start by creating this directories if they don't exist in the
filesystem and adding the proper selinux labels.

bp tls-via-certmonger-containers

Change-Id: Iba3adb9464a755e67c6f87d1233b3affa8be565a

7 years agoEnable encryption of pacemaker traffic by default
Juan Antonio Osorio Robles [Mon, 31 Jul 2017 08:22:22 +0000 (11:22 +0300)]
Enable encryption of pacemaker traffic by default

We already are setting a pre-shared key by default for the pacemaker
cluster. This was done in order to communicate with TLS-PSK with
pacemaker-remote clusters. This key is also useful for us to enable
encrypted traffic for the regular cluster traffic, which we enable by
default with this patch.

Change-Id: I349b8bf79eeeaa4ddde1c17b7014603913f184cf

7 years agoMerge "Enable TLS for the HAProxy stats interface"
Jenkins [Mon, 31 Jul 2017 19:32:33 +0000 (19:32 +0000)]
Merge "Enable TLS for the HAProxy stats interface"

7 years agoUse clustercheck credentials to poll galera state in container
Damien Ciabrini [Mon, 31 Jul 2017 13:10:47 +0000 (09:10 -0400)]
Use clustercheck credentials to poll galera state in container

The clustercheck service currently connects to mysql as root
to poll the state of the galera cluster.

Update the generated config to use clustercheck credentials.

Depends-On: If8e0b3f9e4f317fde5328e71115aab87a5fa655f
Closes-Bug: #1707683

Change-Id: I4ee6e1f56a7880ccf456f5c08d26a267fb810361

7 years agoMerge "Prevent haproxy to run iptables during docker-puppet configuration"
Jenkins [Mon, 31 Jul 2017 15:26:19 +0000 (15:26 +0000)]
Merge "Prevent haproxy to run iptables during docker-puppet configuration"

7 years agoEnable TLS for the HAProxy stats interface
Juan Antonio Osorio Robles [Thu, 29 Jun 2017 12:03:11 +0000 (15:03 +0300)]
Enable TLS for the HAProxy stats interface

This creates a new class for the stats interface and furtherly
configures it to also use the certificates that are provided by
certmonger (via the internal_certificates_specs variable).

Note that the already existing haproxy_stats_certificate still works and
will take precedence if it's set.

bp tls-via-certmonger

Change-Id: Iea65d91648ab13dbe6ec20241a1a7c95ce856e3e

7 years agoUpdate openstackdocstheme>=1.16.0
ZhongShengping [Mon, 31 Jul 2017 03:04:17 +0000 (11:04 +0800)]
Update openstackdocstheme>=1.16.0

Change-Id: I69f9af4191cf7148d517f56c77da739c1a06b49f

7 years agoSupport for Dell EMC Unity Cinder Driver
rajinir [Wed, 26 Jul 2017 19:56:08 +0000 (14:56 -0500)]
Support for Dell EMC Unity Cinder Driver

This changes adds Dell EMC Unity backend as composable service
and matches the tripleo-heat-templates.

Change-Id: I015f7dfec4bedf72332d91b91cda3ef1dc8caf8c

7 years agoHandle SSL options for Zaqar
Thomas Herve [Mon, 24 Jul 2017 11:14:22 +0000 (13:14 +0200)]
Handle SSL options for Zaqar

This allows running Zaqar with SSL under Apache.

Change-Id: I4c68a662c2433398249f770ac50ba0791449fe71

7 years agoPrevent haproxy to run iptables during docker-puppet configuration
Damien Ciabrini [Thu, 20 Jul 2017 15:48:22 +0000 (11:48 -0400)]
Prevent haproxy to run iptables during docker-puppet configuration

When docker-puppet runs module tripleo::haproxy to generate haproxy
configuration file, and tripleo::firewall::manage_firewall is true,
iptables is called to set up firewall rules for the proxied services
and fails due to lack of NET_ADMIN capability.

Make the generation of firewall rule configurable by exposing a
new argument to the puppet module. That way, firewall management can
be temporarily disabled when being run through docker-puppet.

Change-Id: I2d6274d061039a9793ad162ed8e750bd87bf71e9
Partial-Bug: #1697921

7 years agoFix nova and selinux unit tests
Alex Schultz [Thu, 27 Jul 2017 18:54:30 +0000 (12:54 -0600)]
Fix nova and selinux unit tests

The unit tests jobs are failing because of missing pre conditions for
the new shared class introduced by
Ib233689fdcdda391596d01a21f77bd8e1672ae04.  Additionally this change
moved some classes around so that the tests are now failing due to
duplicate class declarations for nova::compute::libvirt::services. This
change moves the include that pulls in the declaration first prior to
the include that exists in tripleo::profile::base::nova::libvirt.

The selinux test was also failing due to a type issue with the fact
being used (boolean vs string)

Change-Id: I5bd4b61d6008820729d58f7743e7e61955dd6f51
Closes-Bug: #1707034

7 years agoMerge "Move gnocchi::api resource to run with wsgi setup"
Jenkins [Thu, 27 Jul 2017 13:36:15 +0000 (13:36 +0000)]
Merge "Move gnocchi::api resource to run with wsgi setup"

7 years agoMerge "Configure redis as incoming storage driver in gnocchi"
Jenkins [Thu, 27 Jul 2017 05:01:07 +0000 (05:01 +0000)]
Merge "Configure redis as incoming storage driver in gnocchi"

7 years agoMove gnocchi::api resource to run with wsgi setup
Juan Antonio Osorio Robles [Mon, 24 Jul 2017 16:22:15 +0000 (19:22 +0300)]
Move gnocchi::api resource to run with wsgi setup

Having this run in step 4 causes a refresh (restart) for httpd, which
in turn is problematic for the gnocchi db upgrade command, since when
it runs httpd is not available at that point. This fixes the issue,
since the API configuration is now ran at the same time as the wsgi
bits.

Change-Id: Ie0ab389a4450bb940757e34d1964423911885fa3

7 years agoPass 'false' docroot to vhost for tls_proxy
Juan Antonio Osorio Robles [Mon, 24 Jul 2017 10:35:56 +0000 (13:35 +0300)]
Pass 'false' docroot to vhost for tls_proxy

passing undef causes a failure since due to a recent commit [1] the
resource now does proper validation of the parameters.

[1] https://github.com/puppetlabs/puppetlabs-apache/commit/d6952b21ec66d7ce8b69dd0c2f2a0debca54e18f

Change-Id: I6dc1e5820a1f4fe449d254d301738e1073f4b82b
Closes-Bug: #1706026

7 years agoConfigure redis as incoming storage driver in gnocchi
Pradeep Kilambi [Wed, 19 Jul 2017 22:39:29 +0000 (18:39 -0400)]
Configure redis as incoming storage driver in gnocchi

puppet support for this is added in Id8d4d091da2611de75390e045ebd473caf2a8909

Change-Id: I3354b54571a1b9d0a9187698217628d273cd7d7e

7 years agoMerge "Puppet module to deploy Manila Share bundle for HA"
Jenkins [Mon, 24 Jul 2017 19:36:48 +0000 (19:36 +0000)]
Merge "Puppet module to deploy Manila Share bundle for HA"

7 years agoMerge "Release pike-3"
Jenkins [Mon, 24 Jul 2017 18:05:37 +0000 (18:05 +0000)]
Merge "Release pike-3"

7 years agoMerge "Deprecates using exec workaround for ODL clustering"
Jenkins [Mon, 24 Jul 2017 17:32:11 +0000 (17:32 +0000)]
Merge "Deprecates using exec workaround for ODL clustering"

7 years agoMerge "Fix lint issues to upgrade to puppet-lint 2.3"
Jenkins [Mon, 24 Jul 2017 17:31:44 +0000 (17:31 +0000)]
Merge "Fix lint issues to upgrade to puppet-lint 2.3"

7 years agoRelease pike-3
Emilien Macchi [Mon, 24 Jul 2017 14:56:44 +0000 (07:56 -0700)]
Release pike-3

Change-Id: I317efb369dc0a6cd4ec9eefb6678d14caba784f9

7 years agoMerge "Make calls to nova::compute::rbd from libvirt profile"
Jenkins [Sat, 22 Jul 2017 08:09:51 +0000 (08:09 +0000)]
Merge "Make calls to nova::compute::rbd from libvirt profile"

7 years agoDeprecates using exec workaround for ODL clustering
Tim Rozet [Fri, 21 Jul 2017 16:59:57 +0000 (12:59 -0400)]
Deprecates using exec workaround for ODL clustering

Previously we had used an exec defined in puppet-tripleo to do
clustering with OpenDaylight docker containers.  The clustering issue is
now fixed in puppet-opendaylight by:
https://git.opendaylight.org/gerrit/#/c/60491

So removing the custom function and class workaround.  Also,
'ha_node_index' is deprecated for configuring clustering with
puppet-opendaylight so that is also removed.

Depends-On: I21c1eb2eff6d4cb855eff4a1122f55ad625d84cc

Change-Id: I7693b692c74071945fdcc08292542e9b458a540b
Signed-off-by: Tim Rozet <trozet@redhat.com>
7 years agoFix lint issues to upgrade to puppet-lint 2.3
Carlos Camacho [Thu, 20 Jul 2017 18:51:14 +0000 (20:51 +0200)]
Fix lint issues to upgrade to puppet-lint 2.3

2017-07-20 15:09:38.571317 | manifests/glance/nfs_mount.pp:65:WARNING: arrow should be on the right operand's line
2017-07-20 15:09:38.571430 | manifests/pacemaker/haproxy_with_vip.pp:107:WARNING: arrow should be on the right operand's line
2017-07-20 15:09:38.571473 | manifests/pacemaker/haproxy_with_vip.pp:108:WARNING: arrow should be on the right operand's line
2017-07-20 15:09:38.571511 | manifests/pacemaker/haproxy_with_vip.pp:109:WARNING: arrow should be on the right operand's line
2017-07-20 15:09:38.571551 | manifests/pacemaker/resource_restart_flag.pp:44:WARNING: arrow should be on the right operand's line
2017-07-20 15:09:38.571590 | manifests/profile/base/cinder/volume/nfs.pp:72:WARNING: arrow should be on the right operand's line
2017-07-20 15:09:38.571625 | manifests/profile/base/docker.pp:188:WARNING: arrow should be on the right operand's line
2017-07-20 15:09:38.571661 | manifests/profile/base/docker.pp:210:WARNING: arrow should be on the right operand's line
2017-07-20 15:09:38.571699 | manifests/profile/base/logging/fluentd.pp:79:WARNING: arrow should be on the right operand's line
2017-07-20 15:09:38.571735 | manifests/profile/base/pacemaker.pp:107:WARNING: arrow should be on the right operand's line
2017-07-20 15:09:38.571773 | manifests/profile/base/swift/ringbuilder.pp:97:WARNING: arrow should be on the right operand's line
2017-07-20 15:09:38.571811 | manifests/profile/base/swift/ringbuilder.pp:125:WARNING: arrow should be on the right operand's line
2017-07-20 15:09:38.571850 | manifests/profile/base/swift/ringbuilder.pp:130:WARNING: arrow should be on the right operand's line
2017-07-20 15:09:38.571889 | manifests/profile/pacemaker/ceph/rbdmirror.pp:79:WARNING: arrow should be on the right operand's line
2017-07-20 15:09:38.571927 | manifests/profile/pacemaker/cinder/backup.pp:66:WARNING: arrow should be on the right operand's line
2017-07-20 15:09:38.571965 | manifests/profile/pacemaker/ovn_northd.pp:96:WARNING: arrow should be on the right operand's line

Change-Id: I9393c5e04310cf84695531df9bb16f33e7e15abb

7 years agoFix up the control-port for rabbitmq bundles
Michele Baldessari [Thu, 20 Jul 2017 17:13:34 +0000 (19:13 +0200)]
Fix up the control-port for rabbitmq bundles

Mistakenly this was set to 3121 which is the same port that pacemaker
remote uses. Move this to 3122 which was the plan all along.

Also fix a wrong port comment in redis and mysql at the same time.

Change-Id: Iccca6a53a769570443091577c7d86f47119d9cbb

7 years agoMerge "PS Cinder: Added support for password less login"
Jenkins [Wed, 19 Jul 2017 21:24:34 +0000 (21:24 +0000)]
Merge "PS Cinder: Added support for password less login"

7 years agoMake calls to nova::compute::rbd from libvirt profile
Giulio Fidente [Wed, 19 Jul 2017 13:19:30 +0000 (15:19 +0200)]
Make calls to nova::compute::rbd from libvirt profile

Some of the tasks carried by nova::compute::rbd class apply libvirt.

Change-Id: Ib233689fdcdda391596d01a21f77bd8e1672ae04
Depends-On: I28557deb13b75922932cd3e86c3467a541c988d0

7 years agoPuppet module to deploy Manila Share bundle for HA
Victoria Martinez de la Cruz [Tue, 11 Jul 2017 20:15:39 +0000 (17:15 -0300)]
Puppet module to deploy Manila Share bundle for HA

This module is used by tripleo-heat-templates to configure and deploy
Kolla-based manila-share containers managed by pacemaker.

We use short-lived containers that call pcs via puppet to create
the needed pacemaker resources, properties and constraints.

Based on work done in fc5bc07b3be401694681420ba453af29b95a9fcf

Change-Id: I89f65e8a34a3a88029498463942016a9f5285f1c
Partial-Bug: #1668922

7 years agoPS Cinder: Added support for password less login
rajinir [Mon, 26 Jun 2017 18:47:04 +0000 (13:47 -0500)]
PS Cinder: Added support for password less login

Added missing san_private_key parameter used for password less SSH
authentication.

Change-Id: Ia9857064692681172573e9092b53a352cd776cbd
Depends-On: 0743d42ed1ed66e08ab7f4355145b4c06c589801

7 years agoMerge "Create a Mesh of qdrouterd links for messaging high availability"
Jenkins [Tue, 18 Jul 2017 05:56:21 +0000 (05:56 +0000)]
Merge "Create a Mesh of qdrouterd links for messaging high availability"

7 years agoMerge "Allow disabling udev usage by LVM"
Jenkins [Tue, 18 Jul 2017 04:38:07 +0000 (04:38 +0000)]
Merge "Allow disabling udev usage by LVM"

7 years agoMerge "Retry ceilometer-upgrade"
Jenkins [Tue, 18 Jul 2017 01:49:41 +0000 (01:49 +0000)]
Merge "Retry ceilometer-upgrade"

7 years agoMerge "Update resource references for dependencies"
Jenkins [Tue, 18 Jul 2017 01:49:25 +0000 (01:49 +0000)]
Merge "Update resource references for dependencies"

7 years agoMerge "Add option for innodb_flush_log_at_trx_commit = 2 for Galera only"
Jenkins [Mon, 17 Jul 2017 08:20:11 +0000 (08:20 +0000)]
Merge "Add option for innodb_flush_log_at_trx_commit = 2 for Galera only"

7 years agoMerge "Modified glance stores parameter values"
Jenkins [Mon, 17 Jul 2017 03:30:21 +0000 (03:30 +0000)]
Merge "Modified glance stores parameter values"

7 years agoCreate a Mesh of qdrouterd links for messaging high availability
John Eckersberg [Tue, 21 Mar 2017 14:05:17 +0000 (10:05 -0400)]
Create a Mesh of qdrouterd links for messaging high availability

For multi-node deployments of the dispatch router, a mesh of
inter-router links is created. Note that bi-directional links must
not be configured.

Example: For nodes A, B, C
  Node    Inter-Router Link
   A:           []
   B:           [A]
   C:           [A,B]

Change-Id: If43beea7a53c1f8f1dff062341c7ea81751c3122

7 years agoRetry ceilometer-upgrade
Alex Schultz [Tue, 11 Jul 2017 21:14:00 +0000 (15:14 -0600)]
Retry ceilometer-upgrade

When the ceilometer-upgrade command is run in step5, it talks to gnocchi
and keystone on all the controllers. Since these other nodes might have
httpd restarted mid-upgrade we should retry if we get a failure.

Change-Id: I874cf9c34b41d055a258704dabe9150eab0f7968
Closes-Bug: #1703444

7 years agoUpdate resource references for dependencies
Emilien Macchi [Thu, 13 Jul 2017 18:26:54 +0000 (11:26 -0700)]
Update resource references for dependencies

The latest version of puppet now reports these as catalog failures so
this change removes the unnecessary references and the references
should be updated.

Closes-Bug: #1702964
Change-Id: Iebc547aa92f9f40e4a633c57d79e6c9cddb5dd28

7 years agoMerge "Add new profile for the Veritas HyperScale's cinder backend."
Jenkins [Sat, 15 Jul 2017 18:33:05 +0000 (18:33 +0000)]
Merge "Add new profile for the Veritas HyperScale's cinder backend."

7 years agoMerge "Contrail: Fix controlplane/dataplane network asignments & enable optional...
Jenkins [Fri, 14 Jul 2017 23:44:24 +0000 (23:44 +0000)]
Merge "Contrail: Fix controlplane/dataplane network asignments & enable optional dpdk"

7 years agoMerge "Fix typo in haproxy bundle"
Jenkins [Fri, 14 Jul 2017 12:44:17 +0000 (12:44 +0000)]
Merge "Fix typo in haproxy bundle"

7 years agoMerge "Add insecure_registry_address to docker profile"
Jenkins [Fri, 14 Jul 2017 10:49:18 +0000 (10:49 +0000)]
Merge "Add insecure_registry_address to docker profile"

7 years agoModified glance stores parameter values
PranaliD [Fri, 14 Jul 2017 07:57:00 +0000 (13:27 +0530)]
Modified glance stores parameter values

The stores parameter should be set with the new parameters
as they are going to be deprecated in the old method.

Change-Id: If272345e96988778ceccb8f2f624db1c38aea365
Closes-Bug: 1704327

7 years agoAdd new profile for the Veritas HyperScale's cinder backend.
abhishek.kane [Tue, 20 Jun 2017 12:23:53 +0000 (17:53 +0530)]
Add new profile for the Veritas HyperScale's cinder backend.

Add new hook in the keystone profile for Veritas HyperScale.
Add new hook in the rabbitmq profile for Veritas HyperScale.
Add new hook in the mysql profile for Veritas HyperScale.

Change-Id: I9168bffa5c73a205d1bb84b831b06081c40af549
Depends-On: I316b22f4f7f9f68fe5c46075dc348a70e437fb1d
Depends-On: Id188af5e2f7bf628a97a70b8f20bef28e42b372d
Signed-off-by: abhishek.kane <abhishek.kane@veritas.com>
Signed-off-by: Dnyaneshwar Pawar <dnyaneshwar.pawar@veritas.com>
7 years agoMerge "Refactor iscsi initiator-name reset into separate profile"
Jenkins [Thu, 13 Jul 2017 23:53:32 +0000 (23:53 +0000)]
Merge "Refactor iscsi initiator-name reset into separate profile"

7 years agoMerge "Fix mysql client config generation with containerized environment"
Jenkins [Thu, 13 Jul 2017 23:38:39 +0000 (23:38 +0000)]
Merge "Fix mysql client config generation with containerized environment"

7 years agoFix typo in haproxy bundle
Michele Baldessari [Thu, 13 Jul 2017 12:18:44 +0000 (14:18 +0200)]
Fix typo in haproxy bundle

Change I6f4d3a5abae8f1781cfe6f69ff960aad500061e3 slipped in a typo
and it removed the '$' character from a puppet manifest. Which causes
a deployment to fail with:
INFO: running container haproxy-bundle-docker-0 for the first time
ERROR: /usr/bin/docker-current: Error response from daemon: Invalid bind mount spec "deployed_ssl_cert_path:deployed_ssl_cert_path:ro": Invalid volume destination path: 'deployed_ssl_cert_path' mount path must be absolute.. See '/usr/bin/docker-current run --help'.
ERROR: docker failed to launch container

Change-Id: Ic602fd443d38482bf1f924531561b2174dc38293

7 years agoAdd insecure_registry_address to docker profile
Dan Prince [Fri, 30 Jun 2017 19:53:54 +0000 (15:53 -0400)]
Add insecure_registry_address to docker profile

This patch adds a new insecure_registry_address parameter
to the docker profile. This parameter is meant to replace two
deprecated parameters which did the same thing.

Co-Authored-By: Ian Main <imain@redhat.com>
Change-Id: I729fa00175cb36b02b882d729aae5ff06d0e3fbc

7 years agoRemove dependency on memcached_node_ips_v6
Steven Hardy [Thu, 6 Jul 2017 11:29:50 +0000 (12:29 +0100)]
Remove dependency on memcached_node_ips_v6

This is set via all_nodes_config in t-h-t, but it's a special case for
this service, so it'll be better if we handle the ipv6 transformation
in puppet instead of relying on the service specific list mangling in
t-h-t (one aspect of which has been identified as a potential performance
problem).

Related-Bug: #1684272
Change-Id: Iccb9089db4b382db3adb9340f18f6d2364ca7f58

7 years agoMerge "Leverage kolla config_files to copy config into containers"
Jenkins [Thu, 13 Jul 2017 01:33:47 +0000 (01:33 +0000)]
Merge "Leverage kolla config_files to copy config into containers"

7 years agoMerge "Refactor nova migration config into client & target profiles"
Jenkins [Wed, 12 Jul 2017 21:14:31 +0000 (21:14 +0000)]
Merge "Refactor nova migration config into client & target profiles"

7 years agoMerge "Let pacemaker bind-mount needed cert for haproxy bundle"
Jenkins [Wed, 12 Jul 2017 19:59:31 +0000 (19:59 +0000)]
Merge "Let pacemaker bind-mount needed cert for haproxy bundle"

7 years agoLeverage kolla config_files to copy config into containers
Martin André [Mon, 26 Jun 2017 13:33:09 +0000 (15:33 +0200)]
Leverage kolla config_files to copy config into containers

This solves a problem with bind-mounts when the containers are holding
files descriptors open.

At the same time this makes the template more robust to puppet changes
since new config files will be available in the containers without
needing to update the templates.

Closes-Bug: #1698323
Change-Id: I857c94ba5f7f064d7c58df621ec5d477654b9166
Depends-On: I78dcec741a941dc21adba33ba33a6dc6ff1d217c

7 years agoFix mysql client config generation with containerized environment
Damien Ciabrini [Mon, 3 Jul 2017 09:15:56 +0000 (05:15 -0400)]
Fix mysql client config generation with containerized environment

When the tripleo::profile::base::database::mysql::client profile is
included by other openstack services, the file /etc/my.cnf.d/tripleo.cnf
is not generated because docker-puppet is configured to disregard the
exec tags.

Make the profile use either File or Exec resource based on how it's
being called, to make it work for both containerized and non-containerized
use cases.

Change-Id: I103baa02373f6713cc300ac039a6f173ff0bbf1c

7 years agoMerge "Do not fail if PCI device is missing"
Jenkins [Wed, 12 Jul 2017 03:28:26 +0000 (03:28 +0000)]
Merge "Do not fail if PCI device is missing"

7 years agoRefactor iscsi initiator-name reset into separate profile
Oliver Walsh [Mon, 10 Jul 2017 14:48:22 +0000 (15:48 +0100)]
Refactor iscsi initiator-name reset into separate profile

This currently assumes nova-compute and iscsid run in the same context which
isn't true for a containerized deployment

Change-Id: I91f1ce7625c351745dbadd84b565d55598ea5b59

7 years agoLet pacemaker bind-mount needed cert for haproxy bundle
Damien Ciabrini [Mon, 10 Jul 2017 12:49:36 +0000 (08:49 -0400)]
Let pacemaker bind-mount needed cert for haproxy bundle

When SSL configuration is enabled, haproxy expects to load a SSL
certificate file at startup.

Update the bundle configuration to always bind-mount the cert
file, to support both SSL and non SSL HAproxy bundle deployments.

Change-Id: I6f4d3a5abae8f1781cfe6f69ff960aad500061e3

7 years agoMerge "Remove reference to bootstack_nodeid"
Jenkins [Sat, 8 Jul 2017 01:06:44 +0000 (01:06 +0000)]
Merge "Remove reference to bootstack_nodeid"

7 years agoMerge "Add Swift dispersion profile"
Jenkins [Thu, 6 Jul 2017 18:12:58 +0000 (18:12 +0000)]
Merge "Add Swift dispersion profile"

7 years agoAdd option for innodb_flush_log_at_trx_commit = 2 for Galera only
Mike Bayer [Mon, 3 Jul 2017 14:35:30 +0000 (10:35 -0400)]
Add option for innodb_flush_log_at_trx_commit = 2 for Galera only

The innodb_flush_log_at_trx_commit flag changes the timing
of when the log buffer is written to disk for writes.
At its default of 1, transactions are written to disk
and the buffer flushed on a per-transaction basis; but when
set to 2, the flush of the buffer proceeds only once per
second.  This removes the durability guarantee for the
single node.  However the central concept of Galera is
that durability is achieved via the cluster as a whole,
in that transactions are replicated to other nodes before
the commit succeeds (though not necessarily written to disk
unless wsrep_causal_reads is set).  In this model,
data would only be lost of all nodes of the Galera cluster
were killed within one second of each other.  Percona's
blog post at https://www.percona.com/blog/2014/11/17/typical-misconceptions-on-galera-for-mysql/
recommends that the value of 2 should be considered "safe"
for a Galera cluster unless you are in fact worried that
all three nodes will be powered off simultaneously.

The value here is added as an option only, defaulting
to the usual default of "1", flush per transaction.

Change-Id: Id5a30f1daf978e094a74db2d284febbc9ae64bb3

7 years agoRemove reference to bootstack_nodeid
Steven Hardy [Thu, 6 Jul 2017 16:28:27 +0000 (17:28 +0100)]
Remove reference to bootstack_nodeid

This has been replaced with bootstrap_nodeid which isn't hard-coded to
the Controller role and thus will work should this service be deployed
on any other role via composable services.

Change-Id: I0a9fced847caf344e5d26b452f1bd40afab8f029

7 years agoContrail: Fix controlplane/dataplane network asignments & enable optional dpdk
Michael Henkel [Fri, 16 Jun 2017 17:49:57 +0000 (10:49 -0700)]
Contrail: Fix controlplane/dataplane network asignments & enable optional dpdk

This patch will move the Contrail roles communication towards
OpenStack APIs from the public/external network to the
internal_api network. I will also add the option to enable
dpdk for Contrail.

Change-Id: Ia835df656031cdf28de20f41ec6ab1c028dced23
Closes-Bug: 1698422

7 years agoMerge "Update puppet-tripleo libvirt to support docker deployments"
Jenkins [Wed, 5 Jul 2017 19:43:53 +0000 (19:43 +0000)]
Merge "Update puppet-tripleo libvirt to support docker deployments"

7 years agoUpdate puppet-tripleo libvirt to support docker deployments
Keith Schincke [Tue, 23 May 2017 18:44:23 +0000 (14:44 -0400)]
Update puppet-tripleo libvirt to support docker deployments

It is not necessary to mangle libvirt_rbd_secret_key parameter as this
is now given by the templates.

Depends-On: Iff3dbcb0f1b4d2373570e184e636a71553cea708
Change-Id: I6b163ab102f505f0d0ce9eb1ad9d4274e4ff6348

7 years agoRefactor nova migration config into client & target profiles
Oliver Walsh [Wed, 7 Jun 2017 22:26:25 +0000 (23:26 +0100)]
Refactor nova migration config into client & target profiles

The nova migration config has always been applied by the base::nova profile.
It assumed that libvirtd/nova-compute and are all running on the
same host.
Where this config didn't apply (e.g a nova api host) it was disabled by a flag.

This approach is not compatible with containers. Hieradata for all containers
are combined so per-host flags no longer work, and we can no longer assume
libvirtd and nova-compute run in the same context.

This change refactors the profiles out of the base nova profile and into
a client profile and a target profile that can be included where appropriate.

Change-Id: I063a84a8e6da64ae3b09125cfa42e48df69adc12
Implements: blueprint tripleo-cold-migration

7 years agoFix openstackdocstheme
ZhongShengping [Sun, 2 Jul 2017 06:28:11 +0000 (14:28 +0800)]
Fix openstackdocstheme

See the docs[0] how to use it. Also add the variables for the bug
log feature.

[0]https://docs.openstack.org/openstackdocstheme/latest/

Change-Id: I11b183986a389291d9ab02cb1d0be36c3d73bdb0

7 years agoMerge "Merge the nova HAproxy TLS options"
Jenkins [Fri, 30 Jun 2017 11:27:06 +0000 (11:27 +0000)]
Merge "Merge the nova HAproxy TLS options"

7 years agoZaqar: support configurable backends
Dan Prince [Mon, 26 Jun 2017 14:24:39 +0000 (10:24 -0400)]
Zaqar: support configurable backends

This patch updates the Zaqar profile so that we have
support for configuring alternate versions of the messaging
and management backends.

In Pike instack-undercloud started using the swift/sqlalchemy
backends and the intent here is to update the new containers
undercloud to use a similar default (thus letting us drop Mongodb).

Change-Id: Ie6a56b9163950cee2c0341afa0c0ddce665f3704

7 years agoDo not fail if PCI device is missing
Brent Eagles [Tue, 27 Jun 2017 12:25:37 +0000 (09:55 -0230)]
Do not fail if PCI device is missing

Fixes a problem where SR-IOV VF count configuration will fail if a
physical function is in use by a guest when 'puppet apply' is executed.
This change substitutes warnings for failures and skips complaints if a
PCI device is unavailable.

Note: this patch has the side-effect of allowing the same configuration
data on hosts that may *not* or *ever* have PCI SR-IOV devices on the
hardware. Time will tell how evil this is in practice.

Closes-Bug: #1701284
Change-Id: I71edc135432ab2193741c37ce977dd11172401e6

7 years agoMerge the nova HAproxy TLS options
Rob Crittenden [Wed, 28 Jun 2017 19:47:20 +0000 (15:47 -0400)]
Merge the nova HAproxy TLS options

This makes sure that we set the necessary options so HAProxy uses TLS
to contact nova.  It was commented out when nova was moved to not run
over httpd. Since that is no longer the case we can re-enable it.

Change-Id: I026a7dab30b00a4e93966f650f098c570b0b624b
Depends-On: Iac35b7ddcd8a800901548c75ca8d5083ad17e4d3

7 years agoMerge "Split docker options and insecure registry"
Jenkins [Wed, 28 Jun 2017 13:50:48 +0000 (13:50 +0000)]
Merge "Split docker options and insecure registry"

7 years agoMerge "Change CRL refresh to run every 2 hours"
Jenkins [Tue, 27 Jun 2017 21:47:13 +0000 (21:47 +0000)]
Merge "Change CRL refresh to run every 2 hours"

7 years agoAllow disabling udev usage by LVM
Jiri Stransky [Tue, 27 Jun 2017 13:34:59 +0000 (15:34 +0200)]
Allow disabling udev usage by LVM

Disabling udev usage from LVM seems to be the only observed working
way of running containerized cinder-volume with local LVM backend.

I didn't come across reports that not using udev would have negative
impact on the functionality.

Additional info at
https://groups.google.com/forum/#!topic/docker-user/n4Xtvsb4RAw

Change-Id: I1bf395a6228dba66fa6bf9b8bcc9f3ac3d922a49
Related-Bug: #1700140

7 years agoSplit docker options and insecure registry
Bogdan Dobrelya [Thu, 22 Jun 2017 11:38:46 +0000 (13:38 +0200)]
Split docker options and insecure registry

Use augeas to modify only parameters' dedicated configuration.
Split options from insecure registry. Overlapping those params may
unschedule the docker service restarts for some cases, ending up with
a split brain state for the docker service run-time config vs changed
/etc/sysconfig/options config.

Change-Id: Ic5640061837b022f7175f0db0dc269f9a61e6023
Signed-off-by: Bogdan Dobrelya <bdobreli@redhat.com>
7 years agoAlways start httpd at the same time (TLS proxy)
Juan Antonio Osorio Robles [Tue, 27 Jun 2017 11:26:57 +0000 (14:26 +0300)]
Always start httpd at the same time (TLS proxy)

For the TLS everywhere job, there are some apache vhosts set up that
serve as TLS proxies. These need to be started at the same time as the
rest of the apache vhosts too.

Change-Id: I15e67c7c04142cff01704e2590d3b2a6a949cc06

7 years agoAlways start httpd at the same time
Juan Antonio Osorio Robles [Wed, 21 Jun 2017 12:43:25 +0000 (15:43 +0300)]
Always start httpd at the same time

Puppet wipes out whatever is not in it's resource catalog each run for
httpd. This causes httpd to restart if in the next step there are
reasources added that were not there earlier.

This patch, thus changes the instances of httpd to start at the same
time: On step 3 for the bootstrap node, and on step 4 for every other
node.

Closes-Bug: #1699502
Change-Id: I3d29728c1ab7bd5b78100f89e00e5fa082f97b0c
Co-Authored-By: Alex Schultz <aschultz@redhat.com>
7 years agoMerge "MySQL: configure ::zaqar::db::mysql selectively"
Jenkins [Tue, 27 Jun 2017 06:02:50 +0000 (06:02 +0000)]
Merge "MySQL: configure ::zaqar::db::mysql selectively"

7 years agoMySQL: configure ::zaqar::db::mysql selectively
Dan Prince [Mon, 26 Jun 2017 14:35:25 +0000 (10:35 -0400)]
MySQL: configure ::zaqar::db::mysql selectively

Adds the ability to create an empty MySQL database for Zaqar
if zaqar is enabled and settings for the mysql backend are
also available in hiera. This should allow Zaqar's database to
get created when needed, but skipped if MongoDB is used
instead (per overcloud defaults).

Change-Id: I3598e39c0a3cdf80b96e728d9aa8a7e6505e0690

7 years agoChange CRL refresh to run every 2 hours
Juan Antonio Osorio Robles [Mon, 19 Jun 2017 08:46:40 +0000 (11:46 +0300)]
Change CRL refresh to run every 2 hours

The default CA issues CRLs for 4 hours by default. So we need to change
these values to reflect this, else we'll get verification issues due to
the CRL having expired before its refreshed.

However, the nextupdate value for the CRLs might not be aligned with the
cron job. And getting this alignment is not entirely trivial. So I opted
for updating every 2 hours to address this.

Change-Id: I732b400462c5cabd7c6c18c007fc9e8c87b700d3

7 years agoForce MySQL users to use SSL if internal TLS is enabled
Juan Antonio Osorio Robles [Fri, 27 Jan 2017 15:42:18 +0000 (17:42 +0200)]
Force MySQL users to use SSL if internal TLS is enabled

This forces the MySQL users to use SSL when connecting to MySQL.

bp tls-via-certmonger

Depends-On: I24e4c195a31109835739e78a6b53d36f661f9fd0
Change-Id: I98856955132b680a159144204da1d5b400fe9794

7 years agoSwitch from oslosphinx to openstackdocstheme
ZhongShengping [Mon, 26 Jun 2017 02:12:30 +0000 (10:12 +0800)]
Switch from oslosphinx to openstackdocstheme

As part of the docs migration work[0] for Pike we need to switch to use the
openstackdocstheme.

[0]https://review.openstack.org/#/c/472275/

Change-Id: Ic82c4a2ec9fe26c8621ab2c7c9598b1582f73156

7 years agoAdd Swift dispersion profile
Christian Schwede [Wed, 21 Jun 2017 19:34:09 +0000 (21:34 +0200)]
Add Swift dispersion profile

The swift-dispersion-populate command needs to be called when Swift and
Keystone are up and running, and therefore we need to ensure this is
running in step 5 or later.

Change-Id: I5b4c08c252b6083dace5a65367920c475de416ce

7 years agoMerge "Remove manifest/profile/baseui.pp"
Jenkins [Fri, 23 Jun 2017 00:30:02 +0000 (00:30 +0000)]
Merge "Remove manifest/profile/baseui.pp"