apex-tripleo-heat-templates.git
6 years agoMerge "Default Ceph pg_num and pgp_num to 128" into stable/pike
Jenkins [Sat, 7 Oct 2017 06:07:37 +0000 (06:07 +0000)]
Merge "Default Ceph pg_num and pgp_num to 128" into stable/pike

6 years agoMerge "Support for Ocata-Pike live-migration over ssh" into stable/pike
Jenkins [Sat, 7 Oct 2017 05:38:24 +0000 (05:38 +0000)]
Merge "Support for Ocata-Pike live-migration over ssh" into stable/pike

6 years agoMerge "Fixes missing type for heat param TenantNetName" into stable/pike
Jenkins [Sat, 7 Oct 2017 05:05:25 +0000 (05:05 +0000)]
Merge "Fixes missing type for heat param TenantNetName" into stable/pike

6 years agoMerge "Use sub_nodes_private instead of node_private" into stable/pike
Jenkins [Sat, 7 Oct 2017 05:00:33 +0000 (05:00 +0000)]
Merge "Use sub_nodes_private instead of node_private" into stable/pike

6 years agoMerge "Update panko port in env ssl yaml files to correct one" into stable/pike
Jenkins [Sat, 7 Oct 2017 03:46:44 +0000 (03:46 +0000)]
Merge "Update panko port in env ssl yaml files to correct one" into stable/pike

6 years agoMerge "Bump fs.inotify.max_user_instances for scale" into stable/pike
Jenkins [Sat, 7 Oct 2017 03:46:38 +0000 (03:46 +0000)]
Merge "Bump fs.inotify.max_user_instances for scale" into stable/pike

6 years agoMerge "Drop extraconfig for nova-nuage" into stable/pike
Jenkins [Sat, 7 Oct 2017 03:46:24 +0000 (03:46 +0000)]
Merge "Drop extraconfig for nova-nuage" into stable/pike

6 years agoMerge "Fixes heat resource name for Internal API Network" into stable/pike
Jenkins [Sat, 7 Oct 2017 03:37:35 +0000 (03:37 +0000)]
Merge "Fixes heat resource name for Internal API Network" into stable/pike

6 years agoSupport for Ocata-Pike live-migration over ssh
Oliver Walsh [Wed, 6 Sep 2017 10:35:07 +0000 (11:35 +0100)]
Support for Ocata-Pike live-migration over ssh

In Ocata all live-migration over ssh is performed on the default ssh port (22).
In Pike the containerized live-migration over ssh is on port 2022 as the
docker host's sshd is using port 22.

To allow live migration during upgrade we need to temporarily pin the Pike
computes to port 22 and in the final converge we can switch over to port 2022.

This also changes the default port to 2022 for baremetal computes in Pike to
enable live-migration between baremetal and containerized computes.

Change-Id: Icb9bfdd9a99dc1dce28eb95c50a9a36bffa621b1
Depends-On: I0b80b81711f683be539939e7d084365ff63546d3
Closes-Bug: 1714171
(cherry picked from commit 17fd16b9f266e1aa67bf03ebdf309e89d668ada2)

6 years agoDefault Ceph pg_num and pgp_num to 128
Giulio Fidente [Thu, 21 Sep 2017 19:18:01 +0000 (21:18 +0200)]
Default Ceph pg_num and pgp_num to 128

As per Ceph docs [1] we should default pg_num and pgp_num to 128 when
using less than 5 OSDs.

This same change was applied to the ceph-ansible profiles with [2].

Also updates the CI environment files to continue using 32 where we
deploy a single OSD.

1. http://docs.ceph.com/docs/master/rados/operations/placement-groups/
2. Ibd9fb23e04576e95e24af58f856663397886a947

Change-Id: I1920bc8f5251f362af38ad3bd6f46dda42c6ee93
Closes-Bug: #1718756
(cherry picked from commit e17ae7620e03790da0d29092ab42e8089b2e8d11)

6 years agoUse sub_nodes_private instead of node_private
Sagi Shnaidman [Mon, 2 Oct 2017 12:33:38 +0000 (15:33 +0300)]
Use sub_nodes_private instead of node_private

node_private file doesn't exist anymore, use sub_nodes_private
instead

Change-Id: Ifb3af18733c0e1fd6895c270bb39199acaa98968

6 years agoFixes missing type for heat param TenantNetName
Tim Rozet [Mon, 2 Oct 2017 15:52:56 +0000 (11:52 -0400)]
Fixes missing type for heat param TenantNetName

Closes-Bug: 1720823

Change-Id: I239cc9f827fe99a553f9c18b80336bc6ce0b1d14
Signed-off-by: Tim Rozet <trozet@redhat.com>
(cherry picked from commit ba5436099d37898e418406f8b4376923e14f4c89)

6 years agoMerge "Pin scenario001-multinode-containers to earlier ceph docker container" into...
Jenkins [Fri, 6 Oct 2017 14:09:17 +0000 (14:09 +0000)]
Merge "Pin scenario001-multinode-containers to earlier ceph docker container" into stable/pike

6 years agoPin scenario001-multinode-containers to earlier ceph docker container
John Fulton [Wed, 4 Oct 2017 17:31:09 +0000 (13:31 -0400)]
Pin scenario001-multinode-containers to earlier ceph docker container

Change-Id: I122a246a559e07ed74c69e3eb172a4bbb801aeb7
Closes-Bug: #1721239
(cherry picked from commit 3e8de70bd5a8c43389432d484189d4de5fc0ae2f)

6 years agoFixes heat resource name for Internal API Network
Tim Rozet [Thu, 21 Sep 2017 19:47:44 +0000 (15:47 -0400)]
Fixes heat resource name for Internal API Network

With the dynamic Jinja2 rendering for networks, the heat resource for
Internal API network was accidentally being renamed to:
OS::TripleO::Network::Internal

when it should be the same as previous versions:
OS::TripleO::Network::InternalApi

This patch removes the 'compat_name' which was overriding the network
name for rendering the resource. This patch also removes the
compat_name functionality from the network/networks.j2.yaml file
since it is no longer needed.

Closes-Bug: 1718764

Change-Id: If756cddd91933edb303cc056515d98b941a3eb14
Signed-off-by: Tim Rozet <trozet@redhat.com>
(cherry picked from commit 97244b942d29d2b5acd7a3eb07acdba0d9b99677)

6 years agoBump fs.inotify.max_user_instances for scale
Sai Sindhur Malleni [Tue, 19 Sep 2017 19:12:35 +0000 (15:12 -0400)]
Bump fs.inotify.max_user_instances for scale

Since each dnsmasq process consumes one inotify socket, the default
value of fs.inotify.max_user_instances which is 128 lets us scale to
only around a 116 neutron subnets (a few other sockets are used by other
processes on the system). Since, we need to provide better defaults,
this patch proposes to bump this value to 1024 by default, while giving
the user a way to cahnge it. Based on
https://unix.stackexchange.com/a/13757 each inotify watch takes 1KB of
memory and we have fs.inotify.max_user_watches set to 8192 by default.
This means that even in the worst case we won't be using more than 8MB
of memory. Bumping the fs.inotify.max_user_instances value to 1024 is
safe because there is fs.inotify.max_user_watches which caps the total
number of files that can be watched by all the inotify instances a user
has.

Related Bugs:
https://bugzilla.redhat.com/show_bug.cgi?id=1474515
https://bugzilla.redhat.com/show_bug.cgi?id=1491505

Change-Id: I39664312bf6cf06f1e1ca2e86ffd86fb9a4582ad
Closes-Bug: 1718266
(cherry picked from commit d2d0c3ff00de9b62382193d942239d543aa9499f)

6 years agoDisable role host_prep_tasks on controlplane upgrade
marios [Tue, 26 Sep 2017 12:33:13 +0000 (15:33 +0300)]
Disable role host_prep_tasks on controlplane upgrade

During the controlplane upgrade the host_prep_tasks are being
executed on the disable_upgrade_deployment roles too.

This sets the role specific host_prep_tasks to an empty list for
those roles during an upgrade, as executing them during the
controlplane upgrade (during -e
major-upgrade-composable-steps-docker.yaml) causes problems.

They will be executed as part of the non controller upgrade as they
are written to the stack outputs to be used as ansible playbooks
(see bug 1708115 for more info on this)

Change-Id: I42c963440b9b1e8222097c3d4e83ffcbe820886c
Closes-Bug: 1719604
(cherry picked from commit 684267a7a4fbff489f6324020289afbdcaaca8f5)

6 years agoMake CephConfigOverrides append to ceph.conf[global]
Giulio Fidente [Wed, 27 Sep 2017 14:39:19 +0000 (16:39 +0200)]
Make CephConfigOverrides append to ceph.conf[global]

Previously it was mistakenly replacing the contents because we
do not do deep merge.

Change-Id: I145feb0208f135da7c71694ebcecd937244d66b1
Closes-Bug: #1719919
(cherry picked from commit 17416dcfc56c5148ccc9ab40297f99adfdcd085b)

6 years agoMerge "Stop mapping docker to OS::Heat::None in scenarios" into stable/pike
Jenkins [Wed, 27 Sep 2017 20:54:34 +0000 (20:54 +0000)]
Merge "Stop mapping docker to OS::Heat::None in scenarios" into stable/pike

6 years agoMerge "Add all services to scenario004-containers" into stable/pike
Jenkins [Wed, 27 Sep 2017 20:38:23 +0000 (20:38 +0000)]
Merge "Add all services to scenario004-containers" into stable/pike

6 years agoMerge "Add a lightweight UC template/role data for deployed-servers" into stable...
Jenkins [Wed, 27 Sep 2017 18:17:28 +0000 (18:17 +0000)]
Merge "Add a lightweight UC template/role data for deployed-servers" into stable/pike

6 years agoMerge "Fix upgrades that use Management network" into stable/pike
Jenkins [Wed, 27 Sep 2017 17:42:37 +0000 (17:42 +0000)]
Merge "Fix upgrades that use Management network" into stable/pike

6 years agoStop mapping docker to OS::Heat::None in scenarios
Jiri Stransky [Wed, 27 Sep 2017 09:15:35 +0000 (11:15 +0200)]
Stop mapping docker to OS::Heat::None in scenarios

This was needed to make the upgrade job on Ocata->Pike passing, and we
now need to remove this to improve the argument order in OOOQ for
deployments with scenarios.

This shouldn't be backported to Ocata (at least not before we make the
split between deploy scenario and upgrade scenario).

Change-Id: Ie08bbe08530bd48a0ca58667f0704f360e0a4dd7
Co-Authored-By: Martin André <m.andre@redhat.com>
Related-Bug: #1714905
Related-Bug: #1712070
(cherry picked from commit 31550b42027588d82f01db6956c1efaf02d58558)

6 years agoAdd all services to scenario004-containers
Jiri Stransky [Tue, 26 Sep 2017 16:31:59 +0000 (18:31 +0200)]
Add all services to scenario004-containers

This commit brings the scenario004 file closer to its BM pendant.  We
need to start with this one to address a chicken-and-egg issue with
featureset files.

Change-Id: Ia5c0cefb7051ca42b4d470f5a000eb446d18be30
Co-Authored-By: Martin André <m.andre@redhat.com>
Related-Bug: #1714905
Related-Bug: #1712070
(cherry picked from commit b4d0a81e55ad51ecdaf2e923f794418ac77cfc57)

6 years agoFixes missing keystone authtoken pw for Tacker
Tim Rozet [Fri, 22 Sep 2017 19:10:42 +0000 (15:10 -0400)]
Fixes missing keystone authtoken pw for Tacker

Closes-Bug: 1718997

Change-Id: I2b347cbc4595e6651b0d4be032cb862fde72e15f
Signed-off-by: Tim Rozet <trozet@redhat.com>
(cherry picked from commit 253d9b9107aa158af5bcdafe510ecd96658ef137)

6 years agoMerge "Move neutron api services to ControllerOpenstack" into stable/pike
Jenkins [Mon, 25 Sep 2017 17:21:00 +0000 (17:21 +0000)]
Merge "Move neutron api services to ControllerOpenstack" into stable/pike

6 years agoMerge "Fix issue where 2 Redis VIPs are assigned, but only one used." into stable...
Jenkins [Mon, 25 Sep 2017 17:20:42 +0000 (17:20 +0000)]
Merge "Fix issue where 2 Redis VIPs are assigned, but only one used." into stable/pike

6 years agoMerge "Fixed resource registry path in neutron-lbaasv2.yaml" into stable/pike
Jenkins [Mon, 25 Sep 2017 17:01:22 +0000 (17:01 +0000)]
Merge "Fixed resource registry path in neutron-lbaasv2.yaml" into stable/pike

6 years agoMerge "Rename service_workflow_tasks into workflow_tasks" into stable/pike
Jenkins [Mon, 25 Sep 2017 14:44:28 +0000 (14:44 +0000)]
Merge "Rename service_workflow_tasks into workflow_tasks" into stable/pike

6 years agoMerge "Remove deploy_steps_tasks.yaml from upgrade_steps_playbook" into stable/pike
Jenkins [Mon, 25 Sep 2017 08:48:30 +0000 (08:48 +0000)]
Merge "Remove deploy_steps_tasks.yaml from upgrade_steps_playbook" into stable/pike

6 years agoFix upgrades that use Management network
Dan Sneddon [Wed, 13 Sep 2017 23:53:36 +0000 (17:53 -0600)]
Fix upgrades that use Management network

Upgrades from older versions using Management network fail.
This patch enables the management network even though it is not
enabled in any of the role definitions. This will allow upgrades
to complete using existing network environment files, without
requiring operators to switch to the new method for defining
which networks are attached to roles. Eventually these older
environment files will be removed.

Change-Id: Iadd12a559f0ad6918958a1355f189187fd327363
Closes-bug: 1717123
(cherry picked from commit 5b9fbc2b2bfa00de2fe0f437f21e05e3fc09a53d)

6 years agoFix issue where 2 Redis VIPs are assigned, but only one used.
Dan Sneddon [Thu, 14 Sep 2017 17:20:54 +0000 (11:20 -0600)]
Fix issue where 2 Redis VIPs are assigned, but only one used.

There is an extra RedisVipPort defined in network-isolation.j2.yaml
which is unused. This will waste an IP address, and can lead to
confusion if there are multiple ports named RedisVipPort.

This patch removes the extra (unneeded) instance of the VIP.

Change-Id: I222873859af1b4ed1050cfffe55687b2f8d4c528
Closes-bug: 1717017
(cherry picked from commit f543752da6e1df3537ffa68d86806e11ac380375)

6 years agoFixed resource registry path in neutron-lbaasv2.yaml
Aneesh Puttur [Wed, 20 Sep 2017 15:13:32 +0000 (11:13 -0400)]
Fixed resource registry path in neutron-lbaasv2.yaml

Change-Id: Icb58d47a3911e83e2650b2c74b33eae522c84651
Closes-Bug: #1718451
(cherry picked from commit edc02b3352d53bdf460a495f689db55944eab432)

6 years agoMerge "Create network-isolation-no-tunneling.yaml using jinja2" into stable/pike
Jenkins [Fri, 22 Sep 2017 21:32:55 +0000 (21:32 +0000)]
Merge "Create network-isolation-no-tunneling.yaml using jinja2" into stable/pike

6 years agoMove neutron api services to ControllerOpenstack
Alex Schultz [Tue, 19 Sep 2017 21:36:43 +0000 (15:36 -0600)]
Move neutron api services to ControllerOpenstack

The Networker role should not have the api services run on it. Instead
these services should run as part of the ControllerOpenstack role that
should be used with this role.

Change-Id: Iabfe276fe700843f3a8da0b9e9220b2f82e20ec9
Closes-Bug: #1718299
(cherry picked from commit 964a5d738b8dbb6beb077d76448c6f3a84be2500)

6 years agoSet Ceph pgp_num after pg_num
Giulio Fidente [Tue, 19 Sep 2017 10:23:17 +0000 (12:23 +0200)]
Set Ceph pgp_num after pg_num

We missed to set the pgp_num default in ceph.conf, causing WARNING
messages like:

  pool default.rgw.buckets.data pg_num 32 > pgp_num 8

Also increases the default pg_num to 128 which is the recommended
value for less than 5 OSDs [1].

1. http://docs.ceph.com/docs/master/rados/operations/placement-groups/

Change-Id: Ibd9fb23e04576e95e24af58f856663397886a947
Closes-Bug: #1718173
(cherry picked from commit 58e6f6533a04eddd2dc897d890737bbccde4ea7b)

6 years agoCreate network-isolation-no-tunneling.yaml using jinja2
Antoni Segura Puimedon [Thu, 31 Aug 2017 09:02:18 +0000 (11:02 +0200)]
Create network-isolation-no-tunneling.yaml using jinja2

The existing network-isolation-no-tunneling.yaml contains
references to missing files. This patch generates the file
with jinja to include custom networks and make it work
with composable networks.

Closes-Bug: #1718797
Change-Id: Ibcab2f6b5ac880a6b3d7dd5126bd24facfa17322
Signed-off-by: Antoni Segura Puimedon <antonisp@celebdor.com>
Co-authored-by: Dan Sneddon <dsneddon@redhat.com>
(cherry picked from commit 47185342bdd247a2e2735ef96c777ecec663086d)

6 years agoMerge "Adds post_upgrade_tasks for any service post-upgrade ansible tasks" into stabl...
Jenkins [Thu, 21 Sep 2017 16:47:21 +0000 (16:47 +0000)]
Merge "Adds post_upgrade_tasks for any service post-upgrade ansible tasks" into stable/pike

6 years agoRemove deploy_steps_tasks.yaml from upgrade_steps_playbook
Marius Cornea [Thu, 14 Sep 2017 11:56:47 +0000 (13:56 +0200)]
Remove deploy_steps_tasks.yaml from upgrade_steps_playbook

After landing https://review.openstack.org/#/c/503484/ we run the
puppet host configuration steps twice. This change removes the
deploy_steps_tasks.yaml playbook in order to run the puppet steps
only once.

Closes-bug: 1717244
Change-Id: I09461094618124915841c8390c8bce8daf64d029
(cherry picked from commit e471c67aab6a8f91011aa2330b3cf80f4427f443)

6 years agoMerge "Make nova patching parameters configurable in Nuage" into stable/pike
Jenkins [Thu, 21 Sep 2017 03:30:35 +0000 (03:30 +0000)]
Merge "Make nova patching parameters configurable in Nuage" into stable/pike

6 years agoMerge "Use haproxy-systemd-wrapper as pid1 in containerized Haproxy" into stable...
Jenkins [Thu, 21 Sep 2017 03:05:45 +0000 (03:05 +0000)]
Merge "Use haproxy-systemd-wrapper as pid1 in containerized Haproxy" into stable/pike

6 years agoMerge "Disable all uses of wsrep-provider in mysql_bootstrap container" into stable...
Jenkins [Thu, 21 Sep 2017 03:05:37 +0000 (03:05 +0000)]
Merge "Disable all uses of wsrep-provider in mysql_bootstrap container" into stable/pike

6 years agoUpdate panko port in env ssl yaml files to correct one
Pradeep Kilambi [Tue, 5 Sep 2017 18:46:52 +0000 (14:46 -0400)]
Update panko port in env ssl yaml files to correct one

Change-Id: Iafe17a91c4695e442881e6fe813a6499f812f4b4
(cherry picked from commit 96667edee266bf2a64f7c8e2488c0eba105eaa8f)

6 years agoUse haproxy-systemd-wrapper as pid1 in containerized Haproxy
Damien Ciabrini [Fri, 15 Sep 2017 11:00:12 +0000 (13:00 +0200)]
Use haproxy-systemd-wrapper as pid1 in containerized Haproxy

This wrapper binary spawns the HAproxy daemon and implements a
coordinated HAproxy restart on SIGHUP.

From a service's perspective, this allows reloading the HAProxy
configuration with minimal service disruption, i.e. without stopping
and restarting the HAProxy container.

Closes-Bug: #1717521

Change-Id: Ib3ef0c0bcf1a8151e179ff4d7509cf0d6b3ac5a1
(cherry picked from commit 91cd44cd7266c15ce07fafbee9d2e33f226096ba)

6 years agoDisable all uses of wsrep-provider in mysql_bootstrap container
Damien Ciabrini [Thu, 14 Sep 2017 12:49:04 +0000 (14:49 +0200)]
Disable all uses of wsrep-provider in mysql_bootstrap container

During the bootstrap of the mariadb database, galera replication
must be disabled while the users credentials are being set up. This
is done by setting wsrep-provider=none when starting mysqld_safe.

Icf67fd2fbf520e8a62405b4d49e8d5169ff3925b already disabled it
when the clustercheck credentials are being set up, but Kolla also
start a temporary server for setting up the root password.

Disable the setting directly at the end of the mysql.cnf in the
running container. That way, the default setting from galera.cnf will
be overriden, all mysqld_safe calls will disable WSREP and the setting
will stay ephemeral.

Change-Id: If14e22992b46a35a05a16a9db5ecb360ea13df8f
Closes-Bug: #1717250
(cherry picked from commit b0f50db80b10e9cd6263c4d6b3ca8dd818b658ba)

6 years agoAdds post_upgrade_tasks for any service post-upgrade ansible tasks
marios [Tue, 15 Aug 2017 13:41:04 +0000 (16:41 +0300)]
Adds post_upgrade_tasks for any service post-upgrade ansible tasks

This adds a new config/deployment per role that will come after any
post deploy steps. It drives the same ansible config as the
upgrade_tasks but instead collects the post_upgrade_tasks for any
service in the given role.

The workflow is upgrade_tasks, then post deploy steps (either
puppet/ or docker/ depending on the env) and then the
post_upgrade_tasks added here.

This is added to the pacemaker/cinder-volume.yaml service for now
see the bug below for more info

Change-Id: Iced34fecf02ebddc91df9302de54d2f4c2cab680
Closes-Bug: 1706951
(cherry picked from commit 2e182bffeeb099cb5e0b1747086fb0e0f57b7b5d)

6 years agoRun gnocchi statsd and metrcd at step 5
Dan Prince [Sat, 26 Aug 2017 02:08:25 +0000 (22:08 -0400)]
Run gnocchi statsd and metrcd at step 5

Running these daemons at step 5 should avoid seeing error messages in
the gnocchi-statsd log files on startup which starts at step4.

Change-Id: Idb82f864a2e1c623dab7a2a87054443036670453
Closes-bug: #1713182
(cherry picked from commit 9d8e496f3e8a825d48d9eba9aab540001bb780ea)

6 years agoChange to boolean for boolean type params
Tong Liu [Tue, 5 Sep 2017 11:27:18 +0000 (11:27 +0000)]
Change to boolean for boolean type params

Some boolean params are set to string type. Although it works, but
it is better to use boolean type for better validation. This patch
changes them to boolean type.

Change-Id: I9f1d223619ea14fbab26033b24eb1144796e5ef2
Closes-Bug: #1715209
(cherry picked from commit cab8ab1d342c6ffada3f2adea5834b4549240af5)

6 years agoOne time delete pacemaker resources during upgrade to containers
Marius Cornea [Thu, 7 Sep 2017 13:38:54 +0000 (15:38 +0200)]
One time delete pacemaker resources during upgrade to containers

This change allows running the major upgrade composable docker
steps multiple times by not trying to delete the pacemaker resources
if they're not reported as started or in master state.

Closes-bug: 1716031
Depends-On: I8da03f5c4a6d442617b81be5793a9724cc8842bf
Change-Id: Ifcf9de8c82550a90a9fb118052d43fdbcdc6ca7e
(cherry picked from commit 64d7be1e3d4552e06cbc53f788572e530cc5c3bb)

6 years agoMake nova patching parameters configurable in Nuage
Lokesh Jain [Wed, 23 Aug 2017 22:12:37 +0000 (18:12 -0400)]
Make nova patching parameters configurable in Nuage

Nova patching parameters are available in nova.conf but are not
configurable from tripleo-heat-templates. Exposing these parameters
from Nuage composable services to make them configurable. It enables
setting the patching parameters in environment files. This change
depends on the addition of nova patching configuration parameters.

Change-Id: Iacad25da044f2bac83ee5f577ddcd70650eb61e5
Depends-On: I51ef3e19daff1d98cfe5c2c16475c16e6a3e3e0f
(cherry picked from commit f0041153eca8d82bb7f72dc68676cab8448ef037)

6 years agoRename service_workflow_tasks into workflow_tasks
Giulio Fidente [Tue, 12 Sep 2017 20:29:13 +0000 (22:29 +0200)]
Rename service_workflow_tasks into workflow_tasks

Using the service_ prefix seems incoherent with its use in
service_config_settings (vs config_settings).

Change-Id: Ia39f181415bee0071409dabddfa0c5c312915e1f
(cherry picked from commit 09137304b98a02ed024c0288da907cfe35ca5fe1)

6 years agoRetry if the pacemaker_resource commands failed
Mathieu Bultel [Mon, 28 Aug 2017 15:24:47 +0000 (17:24 +0200)]
Retry if the pacemaker_resource commands failed

Add a retry when the pacemaker_resource command
wasn't apply correctly, more info here:
https://bugzilla.redhat.com/show_bug.cgi?id=1482116

This is the same approach puppet-pacemaker uses
and provides eventual consistency when multiple
nodes change the cluster CIB concurrently.

This change depends-on :
https://review.gerrithub.io/375982

The return code is not available in the current
ansible-pacemaker package.

Change-Id: I8da03f5c4a6d442617b81be5793a9724cc8842bf
(cherry picked from commit e92430d8d03fc2ce2d0ce192b96209f2c5c04169)

6 years agoMerge "Enable redis TLS proxy in HA deployments" into stable/pike
Jenkins [Wed, 13 Sep 2017 22:34:26 +0000 (22:34 +0000)]
Merge "Enable redis TLS proxy in HA deployments" into stable/pike

6 years agoMerge "Add CephConfigOverrides to allow arbitrary configs in ceph.conf" into stable...
Jenkins [Wed, 13 Sep 2017 21:26:58 +0000 (21:26 +0000)]
Merge "Add CephConfigOverrides to allow arbitrary configs in ceph.conf" into stable/pike

6 years agoMerge "Add RoleConfig output to major_upgrade_steps.j2.yaml" into stable/pike
Jenkins [Wed, 13 Sep 2017 17:46:35 +0000 (17:46 +0000)]
Merge "Add RoleConfig output to major_upgrade_steps.j2.yaml" into stable/pike

6 years agoMerge "Enable selinux in containers" into stable/pike
Jenkins [Wed, 13 Sep 2017 04:57:05 +0000 (04:57 +0000)]
Merge "Enable selinux in containers" into stable/pike

6 years agoMerge "Add verbose output to containerized cell_v2 host discovery" into stable/pike
Jenkins [Wed, 13 Sep 2017 04:55:53 +0000 (04:55 +0000)]
Merge "Add verbose output to containerized cell_v2 host discovery" into stable/pike

6 years agoAdd RoleConfig output to major_upgrade_steps.j2.yaml
Steven Hardy [Mon, 11 Sep 2017 13:39:06 +0000 (14:39 +0100)]
Add RoleConfig output to major_upgrade_steps.j2.yaml

I96ec09bc788836584c4b39dcce5bf9b80e914c71 added this output to the
deploy-steps.j2, but missed adding this to the major upgrade template
which means the overcloud RoleConfig output is broken after the upgrade
(until the converge update switches back to the deploy-steps.j2 derived
template)

Closes-Bug: #1716404
Change-Id: I331fa18b456ca2d6c124316d513374e3fe5a5007
(cherry picked from commit 27018b4182d77abf612697cfe54a4fc3ceeb6be5)

6 years agoDrop extraconfig for nova-nuage
Vineet Paul [Wed, 16 Aug 2017 09:49:40 +0000 (05:49 -0400)]
Drop extraconfig for nova-nuage

Made the Compute as a composable service with Nuage.
Moved all the Nuage specific parameters from extraconfig to be part of this service.

Change-Id: Ic83e9c18d09fbba62bb5d8a12e28a23127f4197d
(cherry picked from commit 4b1276b8f6fec22ac3764d58c4ef647535c85cb9)

6 years agoAdd CephConfigOverrides to allow arbitrary configs in ceph.conf
Giulio Fidente [Wed, 6 Sep 2017 07:39:12 +0000 (09:39 +0200)]
Add CephConfigOverrides to allow arbitrary configs in ceph.conf

We need to reuse the ceph_conf_overrides structure provided by
ceph-ansible for both user provided configs and TripleO managed
configs. This change merges the special user facing parameter
with the TripleO generated configs.

Also adds osd_scenario and osd_objectstore params for compatibility
with newer ceph-ansible versions.

Change-Id: I29c689c6c689590da5b6a3f581fdbec98a52e207
Closes-Bug: #1715321
(cherry picked from commit 32bc2abf14af4ca1449e18b848e2be3cff013987)

6 years agoMerge "Add panko config to ceilometer notification agent container" into stable/pike
Jenkins [Tue, 12 Sep 2017 04:22:36 +0000 (04:22 +0000)]
Merge "Add panko config to ceilometer notification agent container" into stable/pike

6 years agoMerge "Fixes OpenDaylight updating port status" into stable/pike
Jenkins [Tue, 12 Sep 2017 04:22:29 +0000 (04:22 +0000)]
Merge "Fixes OpenDaylight updating port status" into stable/pike

6 years agoMerge "Add a docker pull retry to docker-puppet.py" into stable/pike
Jenkins [Tue, 12 Sep 2017 04:22:22 +0000 (04:22 +0000)]
Merge "Add a docker pull retry to docker-puppet.py" into stable/pike

6 years agoMerge "Add DhcpAgentNotification param to neutron base" into stable/pike
Jenkins [Tue, 12 Sep 2017 04:21:38 +0000 (04:21 +0000)]
Merge "Add DhcpAgentNotification param to neutron base" into stable/pike

6 years agoMerge "Persist containerized services httpd logs" into stable/pike
Jenkins [Tue, 12 Sep 2017 04:20:52 +0000 (04:20 +0000)]
Merge "Persist containerized services httpd logs" into stable/pike

6 years agoEnable selinux in containers
Oliver Walsh [Tue, 5 Sep 2017 18:19:17 +0000 (19:19 +0100)]
Enable selinux in containers

We cannot use the --selinux-enabled docker daemon option on CentOS/RHEL 7.3.
It will fail if security_inode_copy_up is not found in the kernel symbols:
https://github.com/projectatomic/docker/blob/docker-1.12.6/daemon/daemon_unix.go#L661
NB this has been reduced to a warning upstream:
https://github.com/moby/moby/commit/885b29df096db1d6746ece4b3a298a1ffe85716d

Instead this just bind mounts /sys/fs/selinux in containers-common.yaml.

Everything appears to work at initial glance. Pingtest succeeds, and
live-migration between baremetal and containerized computes works.

Change-Id: I018221bf7ae9ab9ece193b55f1ce31eb1591046c
Closes-bug: #1715171
(cherry picked from commit 520f889a31f1ea6ee2bad86d1dbb3c0435604d10)

6 years agoAdd verbose output to containerized cell_v2 host discovery
Oliver Walsh [Fri, 25 Aug 2017 15:11:24 +0000 (16:11 +0100)]
Add verbose output to containerized cell_v2 host discovery

Required to debug issues.

Change-Id: I4d86c8d9ecc353a916475977eb6f2d842c812556
(cherry picked from commit dc64a1108e7bc23f92d77e75001fb42549731e3b)

6 years agoAdd panko config to ceilometer notification agent container
Pradeep Kilambi [Wed, 6 Sep 2017 13:03:37 +0000 (09:03 -0400)]
Add panko config to ceilometer notification agent container

Without this, ceilometer notification agent cant find panko
and skips posting events to it.

Change-Id: Ibfeef5c557d1ceb11a999aa947597014ca94ec34
(cherry picked from commit 5437086ee744469b9daf8cd9edd600f7aa98dde6)

6 years agoEnable redis TLS proxy in HA deployments
Martin André [Mon, 28 Aug 2017 14:50:28 +0000 (16:50 +0200)]
Enable redis TLS proxy in HA deployments

Redis does not have TLS out of the box. Let's use a proxy container for
TLS termination.

This commit enables redis TLS proxy for the HA deployment.

bp tls-via-certmonger

Change-Id: I45e539872a03878337def33c681c4577c1a5629e
(cherry picked from commit c6d8df01d7aa8b44af9ac152b3bb08f07e2e02b7)

6 years agoMerge "Add defaults for ceilometer-agent-compute upgrade tasks" into stable/pike
Jenkins [Mon, 11 Sep 2017 17:49:26 +0000 (17:49 +0000)]
Merge "Add defaults for ceilometer-agent-compute upgrade tasks" into stable/pike

6 years agoMerge "Enable Ceilometer agent logging for containers" into stable/pike
Jenkins [Mon, 11 Sep 2017 17:47:47 +0000 (17:47 +0000)]
Merge "Enable Ceilometer agent logging for containers" into stable/pike

6 years agoAdd DhcpAgentNotification param to neutron base
Tong Liu [Wed, 30 Aug 2017 17:53:33 +0000 (17:53 +0000)]
Add DhcpAgentNotification param to neutron base

Add DhcpAgentNotification param in neutron base yaml file to allow
user to toggle dhcp_agent_notification for neutron.

Change-Id: I31715f58e885ac0c1cd9d813f79df9906b780d99
Closes-Bug: #1713193
(cherry picked from commit 5ea728cba456f3833a626f86043f17427bca5d4f)

6 years agoMerge "Add Neutron SR-IOV agent container" into stable/pike
Jenkins [Mon, 11 Sep 2017 15:57:52 +0000 (15:57 +0000)]
Merge "Add Neutron SR-IOV agent container" into stable/pike

6 years agoMerge "Disables QoS with OpenDaylight deployments" into stable/pike
Jenkins [Mon, 11 Sep 2017 15:40:15 +0000 (15:40 +0000)]
Merge "Disables QoS with OpenDaylight deployments" into stable/pike

6 years agoPersist containerized services httpd logs
Bogdan Dobrelya [Wed, 2 Aug 2017 09:44:18 +0000 (11:44 +0200)]
Persist containerized services httpd logs

Store the httpd logs under dedicated /var/log/containers/httpd/
paths.
Additionally, add release notes describing upgrade impact
for containerized services logs.

Closes-bug: #1700045

Change-Id: I8120c56f2315700862bd0f708b8baa8910275b09
Signed-off-by: Bogdan Dobrelya <bdobreli@redhat.com>
(cherry picked from commit 287e84585ca9170570ce8d06eebd7f9a3ec3345c)

6 years agoAdd a docker pull retry to docker-puppet.py
Dan Prince [Thu, 7 Sep 2017 20:48:28 +0000 (16:48 -0400)]
Add a docker pull retry to docker-puppet.py

Co-Authored-By: Ian Main <imain@redhat.com>
Change-Id: Iad6d38690340f4a064a4527c58ed439d91fa5188
Closes-bug: #1715136
(cherry picked from commit d3b3361a76c2e8b188fa8e586d9fb7f3c60bb66f)

6 years agoEnable Ceilometer agent logging for containers
Pradeep Kilambi [Wed, 6 Sep 2017 16:14:12 +0000 (12:14 -0400)]
Enable Ceilometer agent logging for containers

Change-Id: Ibeb28d7c497b02253d00a74257989cefba2b0cc4
(cherry picked from commit fc44ee6ff3553754c618349df3be7544b17e9c5f)

6 years agoAdd defaults for ceilometer-agent-compute upgrade tasks
Marius Cornea [Thu, 7 Sep 2017 08:56:29 +0000 (10:56 +0200)]
Add defaults for ceilometer-agent-compute upgrade tasks

This change allows the upgrade non controller script, which loops
throug all steps, to complete by adding default values to be
evaluated in the steps where the vars are not registered.

Closes-Bug: 1715574

Change-Id: Ic056fc556240d1acc9f28a75f63c7628cc64da03
(cherry picked from commit d109c1d7a7d2f6302c39369de8a601bc0b2f6704)

6 years agoMerge "Mount vhost_sockets directory for vhost-user socket creation" into stable...
Jenkins [Mon, 11 Sep 2017 11:50:15 +0000 (11:50 +0000)]
Merge "Mount vhost_sockets directory for vhost-user socket creation" into stable/pike

6 years agoMove the clustercheck service to the DB role
Michele Baldessari [Fri, 8 Sep 2017 10:31:18 +0000 (12:31 +0200)]
Move the clustercheck service to the DB role

The clustercheck service is currently in the ControllerOpenstack role
which represents a controller without the DB. Since the clustercheck
service/container always talks to the SQL server via a localhost
connection it *has* to run on the very same node that hosts the DB.

In a containerized deployment this error shows up with db syncs simply
hanging because haproxy will stop serving port 3306 because the
clustercheck service on port 9200 cannot talk to mysql locally.

Errors like this will be logged when trying to connect to the DB VIP:
mysql -u heat -h 172.17.1.13 -p3UazsaeTC64V9UvEcJ3GZ9rbd
ERROR 2013 (HY000): Lost connection to MySQL server at 'reading initial communication packet', system error: 0

Fix this by making sure that the clustercheck service runs on
the DB role.

Change-Id: Iec4c9678d8b8d44e002c1e53110dedc0674359fb
Closes-Bug: #1715847
(cherry picked from commit 1760079dfe5905f2e696b9fc5c729cffa44554ae)

6 years agoDisables QoS with OpenDaylight deployments
Itzik Brown [Wed, 9 Aug 2017 08:00:35 +0000 (11:00 +0300)]
Disables QoS with OpenDaylight deployments

QoS is not fully supported and fails to load correctly with
networking-odl.  Therefore disabling it from Neutron extension drivers
until we have it fully working.

Change-Id: I89aa3628c1400305f9659f5c0c99942a7fa7d19e
Closes-Bug: 1708131
(cherry picked from commit cfd0d185a93ac2922e233e268a32c3574bee37bf)

6 years agoFixes OpenDaylight updating port status
Tim Rozet [Wed, 6 Sep 2017 20:30:42 +0000 (16:30 -0400)]
Fixes OpenDaylight updating port status

ODL now uses a websocket port to update the port status to Neutron.
This port (8185) was blocked so port updates were never received in
Neutron and instances would not come up properly.  This patch opens the
port for ODL deployments.

Closes-Bug: 1715484

Change-Id: Ic59b224c67c02b56b0273700e8e2aa85ae6f8c88
Signed-off-by: Tim Rozet <trozet@redhat.com>
(cherry picked from commit e2558c4a665345e67fcc784c21188bdf06ff1126)

6 years agoMount vhost_sockets directory for vhost-user socket creation
Saravanan KR [Wed, 30 Aug 2017 09:14:53 +0000 (14:44 +0530)]
Mount vhost_sockets directory for vhost-user socket creation

For DPDK, vhost-user sockets are created on the host at
/var/lib/vhost_sockets directory, which will be used by
libvirt and openvswitch. This directory has the necessary
permissions and SELinux policies. Mount this folder for
libvirt container.

Change-Id: Id8be208d1b05886ac45dfdcf48fe766ee5724d1c
Partial-Bug: #1712732
(cherry picked from commit 3ea04744c22ae4cd2e1f2b77fc7d5ade012899e0)

6 years agoMerge "Maintain ceph-osd package only on nodes hosting CephOSD service" into stable...
Jenkins [Fri, 8 Sep 2017 06:22:09 +0000 (06:22 +0000)]
Merge "Maintain ceph-osd package only on nodes hosting CephOSD service" into stable/pike

6 years agoMerge "Add tags in upgrade_tasks for mongodb services." into stable/pike
Jenkins [Thu, 7 Sep 2017 19:31:40 +0000 (19:31 +0000)]
Merge "Add tags in upgrade_tasks for mongodb services." into stable/pike

6 years agoMerge "Add CephExternal role for ceph-ansible" into stable/pike
Jenkins [Thu, 7 Sep 2017 19:31:32 +0000 (19:31 +0000)]
Merge "Add CephExternal role for ceph-ansible" into stable/pike

6 years agoMerge "Support HA for OVN DBs containers using pacemaker bundle" into stable/pike
Jenkins [Thu, 7 Sep 2017 18:32:04 +0000 (18:32 +0000)]
Merge "Support HA for OVN DBs containers using pacemaker bundle" into stable/pike

6 years agoMerge "Use containerized mongodb in scenario002-multinode-containers" into stable...
Jenkins [Thu, 7 Sep 2017 17:32:34 +0000 (17:32 +0000)]
Merge "Use containerized mongodb in scenario002-multinode-containers" into stable/pike

6 years agoMerge "fluentd: log configuration was not generated correctly" into stable/pike
Jenkins [Thu, 7 Sep 2017 16:39:47 +0000 (16:39 +0000)]
Merge "fluentd: log configuration was not generated correctly" into stable/pike

6 years agoMerge "Add support for deploying RGW with ceph-ansible" into stable/pike
Jenkins [Thu, 7 Sep 2017 13:14:02 +0000 (13:14 +0000)]
Merge "Add support for deploying RGW with ceph-ansible" into stable/pike

6 years agoAdd tags in upgrade_tasks for mongodb services.
Jose Luis Franco Arza [Wed, 6 Sep 2017 09:50:00 +0000 (11:50 +0200)]
Add tags in upgrade_tasks for mongodb services.

Patch Ie09ce2a52128eef157e4d768c1c4776fc49f2324 added a new
set of upgrade tasks which were missing the 'tags' keyword.

Closes-Bug: 1715631
Change-Id: Ib1c1aadfbf58c9bccc18667934c8b3c5f38fafa4
(cherry picked from commit 7897d38274cb6435289bc4f4928f96b111e5b4f4)

6 years agoAdd Neutron SR-IOV agent container
Brent Eagles [Tue, 30 May 2017 11:52:55 +0000 (09:22 -0230)]
Add Neutron SR-IOV agent container

This patch adds support for running the neutron SR-IOV agent in a
container.

Depends-On: I4a63845a97c890d7d408731ec5509c320289f18f
Depends-On: Ie5d8cd7863c0d042cc6a4e1fc52602d8a03a1935
Depends-On: I1b5ab0a64ae1f5735f1bd5a68e6ae8bdcf47ddec

Closes-Bug: #1715388

Change-Id: I7ee603b32eddacd02d846dff00dd1b786d4a7ad9
(cherry picked from commit 94c9c2f954e85de0ab895926a969587b90bc4191)

6 years agoAdd CephExternal role for ceph-ansible
Giulio Fidente [Thu, 31 Aug 2017 14:42:16 +0000 (16:42 +0200)]
Add CephExternal role for ceph-ansible

Previously it was only possible to configure the overcloud with
an external Ceph cluster via puppet-ceph-external.

This submission adds a CephExternal implementation which uses
ceph-ansible.

Change-Id: Id0d375f88e27e91e9d89f25a0cd7388b6e45df8b
Depends-On: Ifc57c9cf6ca8017a2abc78d6320c0675ad49ca9f
Closes-Bug: #1714271
(cherry picked from commit 01e55c314de74579196518d958bf5be30e390409)

6 years agoMerge "Use DeployedSSLCertificatePath for public TLS via certmonger" into stable...
Jenkins [Thu, 7 Sep 2017 11:32:18 +0000 (11:32 +0000)]
Merge "Use DeployedSSLCertificatePath for public TLS via certmonger" into stable/pike

6 years agoAdd support for deploying RGW with ceph-ansible
Keith Schincke [Thu, 6 Jul 2017 02:16:26 +0000 (22:16 -0400)]
Add support for deploying RGW with ceph-ansible

This patch allows usage of ceph-ansible to configure the RGW service
in the overcloud. Still uses puppet-keystone to create the necessary
user and endpoint in the catalog.

Co-Authored-By: Giulio Fidente <gfidente@redhat.com>
Change-Id: Iafa17bb64c54e40350b2ba7d76dea3d82fcab0e4
(cherry picked from commit 5b3cd1dcacff408bcb482bdea6cded8755a39ebb)

6 years agoMerge "Containerized mongodb, disable by default, fix upgrade" into stable/pike
Jenkins [Thu, 7 Sep 2017 05:34:35 +0000 (05:34 +0000)]
Merge "Containerized mongodb, disable by default, fix upgrade" into stable/pike

6 years agoMerge "Change all references of nsx_v3 to nsx." into stable/pike
Jenkins [Thu, 7 Sep 2017 05:34:28 +0000 (05:34 +0000)]
Merge "Change all references of nsx_v3 to nsx." into stable/pike

6 years agoUse DeployedSSLCertificatePath for public TLS via certmonger
Juan Antonio Osorio Robles [Mon, 4 Sep 2017 11:04:28 +0000 (14:04 +0300)]
Use DeployedSSLCertificatePath for public TLS via certmonger

As described in the bug report, DeployedSSLCertificatePath is used by
the TLS injection script (if you decide to use that).

There is an alternative, which is to use FreeIPA to provide the
certificate for public TLS (powered by certmonger); however, it doesn't
use the same path as what folks expected. This reuses the
DeployedSSLCertificatePath parameter and uses that as a path for the
resulting PEM file, so its easier to debug.

Change-Id: If73c9599d8b94d2f02b8e4c48f4a235e0fea764d
Closes-Bug: #1714932
(cherry picked from commit f395d9eab2277061e926f7956bb3a56b0c7b1131)

6 years agoMaintain ceph-osd package only on nodes hosting CephOSD service
Alan Bishop [Wed, 30 Aug 2017 13:26:16 +0000 (09:26 -0400)]
Maintain ceph-osd package only on nodes hosting CephOSD service

The ceph-osd package is only required on nodes hosting the CephOSD
service, but the package's presence on other nodes may interfere with
software updates. That's because some distros distribute Ceph software
in different channels, and not all nodes have access to the ceph-osd
channel.

There are two parts to the fix, and the first is an enhancement to the
yum update process. The process detects when the ceph-osd package is not
required, and removes the package from the node.

The second part takes ceph-osd out of the default list of packages
needed by puppet-ceph. The ceph-osd package is listed only on the nodes
hosting the CephOSD service.

Closes-Bug: #1713292
Change-Id: I7a581518ed25cf5f264abfaabfcf2041363a065b
(cherry picked from commit 5a89ea21f2add98119a10464b020a98999d31c41)

6 years agofluentd: log configuration was not generated correctly
Lars Kellogg-Stedman [Sat, 26 Aug 2017 04:36:44 +0000 (21:36 -0700)]
fluentd: log configuration was not generated correctly

fluentd hiera elements were being set in all_nodes.json, but then were
overwritten by values in <role>.json (e.g., controller.json). This
commit removes the values from all-nodes.json and ensures that they
are set correctly in <role>.json.

Closes-Bug: #1713240
Change-Id: I2b4c74c2a807f8e2fed57112f06b3791701bbe95
(cherry picked from commit d9db0c5f4f0fb07832e54b1c7fd7f5c8bfd4134e)