fuel.git
6 years ago[patch] Use keystoneclient to check project ID 73/49873/2
Alexandru Avadanii [Wed, 3 Jan 2018 00:03:42 +0000 (01:03 +0100)]
[patch] Use keystoneclient to check project ID

Another prerequisite for decoupling public network from Openstack
internal management network (upstream won't fix it for Pike):
- port fix from [1] for using the internal network when connecting
  to keystone during project ID validation in nova, instead of
  going through public endpoint (and using SSL).

[1] https://bugs.launchpad.net/nova/+bug/1716344

Change-Id: Ic9a307df9af78fcd58cbcc07b5e62a7e07cc8d7d
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
6 years ago[patch] keystone: Use v3 for admin endpoint 43/49843/3
Alexandru Avadanii [Tue, 2 Jan 2018 04:45:42 +0000 (05:45 +0100)]
[patch] keystone: Use v3 for admin endpoint

Now that v2 API is obsolete, also switch 'admin' endpoint to v3
(previously it was kept back for OCL compatibility).

Change-Id: I9775d59d5e6b93d7351157f7550a0dd7114bee2f
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
6 years agoMerge "[baremetal] heat: Switch metadata API URL to mgmt"
Alexandru Avadanii [Tue, 2 Jan 2018 19:06:00 +0000 (19:06 +0000)]
Merge "[baremetal] heat: Switch metadata API URL to mgmt"

6 years agoMerge "[baremetal] prx: Add management network VIP"
Alexandru Avadanii [Tue, 2 Jan 2018 19:04:40 +0000 (19:04 +0000)]
Merge "[baremetal] prx: Add management network VIP"

6 years ago[patch] Fix OVS ifup workaround 71/49871/1
Alexandru Avadanii [Tue, 2 Jan 2018 18:11:24 +0000 (19:11 +0100)]
[patch] Fix OVS ifup workaround

Do not assume routes are on the same OVS port as the one currently
being configured. Instead, apply the `unless` ifup condition for
any OVS port.

Change-Id: Iea8084f9e50401d300feb7ed16f90b430680cac5
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
6 years ago[baremetal] heat: Switch metadata API URL to mgmt 41/49841/1
Alexandru Avadanii [Tue, 2 Jan 2018 04:33:19 +0000 (05:33 +0100)]
[baremetal] heat: Switch metadata API URL to mgmt

Prepare for decoupling management from public (drop mas01 NAT):
- ctl: change heat URLs to use new management VIP instead of public;

Change-Id: I8e220ee37bd4177c3afd58a9ee401f815d046706
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
6 years ago[baremetal] prx: Add management network VIP 33/49833/2
Alexandru Avadanii [Tue, 2 Jan 2018 04:30:39 +0000 (05:30 +0100)]
[baremetal] prx: Add management network VIP

Include `openstack_web_public_vip` class for setting up the
old VIP in the public network, use old class for mgmt VIP.
Also change the generic hostname 'prx' to point inside mgmt net.

Change-Id: Iff69394f16ede290d149a26b054a85371f00f8e0
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
6 years agodocs: diagrams: Add PXE/admin on cfg01 31/49831/1
Alexandru Avadanii [Mon, 1 Jan 2018 19:42:20 +0000 (21:42 +0200)]
docs: diagrams: Add PXE/admin on cfg01

Refresh diagrams to reflect that the internal network is not used
anymore on jumpserver after PXE/admin was hooked to cfg01.

Change-Id: I4c162d59824e182bc76c0a395742050544e95291
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
6 years ago[baremetal] MaaS: Enable HTTP proxy 57/49157/10
Alexandru Avadanii [Sun, 17 Dec 2017 18:34:18 +0000 (19:34 +0100)]
[baremetal] MaaS: Enable HTTP proxy

Instead of using NAT on the mas01 node for all cluster node outgoing
traffic, use the MaaS built-in proxy for APT traffic to leverage its
caching capabilities too.
Also enable the proxy for salt minions, so they can access public
keyservers et al.
Cleanup public DNS from kvm nodes, interferes with MaaS proxy.

Add example config for global env proxy, but don't enable it:
- default environment settings - /etc/environment (via reclass);

The MaaS proxy will not be used (at least for now) on nodes:
- cfg01;
- mas01;

NOTE: We can't yet drop the maas.pxe_nat state completely, as certain
Openstack services are still accessed via public addresses from ctl
nodes.

JIRA: FUEL-317
JIRA: FUEL-318

Change-Id: I6c5f6872bb94afb838580571080e808bc262fc68
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
6 years ago[baremetal] cmp: Add missing public gw (default) 91/49791/2
Alexandru Avadanii [Sat, 30 Dec 2017 22:04:17 +0000 (23:04 +0100)]
[baremetal] cmp: Add missing public gw (default)

When we dropped the default gw via mas01 NAT, we uncovered a bug,
compute nodes do not have the proper public gw set up and used
to reach public network via mas01, slowing everything down.
Add gw similar to prx nodes.

Fixes: d4ab072

Change-Id: I4343c31c376a7a223670cdd623366454396d8d92
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
6 years agosalt: Use apt-mk 'stable' distribution 77/49777/1
Alexandru Avadanii [Sat, 30 Dec 2017 15:18:41 +0000 (16:18 +0100)]
salt: Use apt-mk 'stable' distribution

'nightly' repo dist from apt-mk is broken, so switch to 'stable'.

Change-Id: Ie12dfc2a499910b8b98a63886ba16044e66435f5
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
6 years agoMerge "[netconfig] APT: Prefer ipv4 connections over ipv6"
Alexandru Avadanii [Fri, 29 Dec 2017 13:22:21 +0000 (13:22 +0000)]
Merge "[netconfig] APT: Prefer ipv4 connections over ipv6"

6 years ago[netconfig] APT: Prefer ipv4 connections over ipv6 33/49733/2
Alexandru Avadanii [Thu, 28 Dec 2017 14:32:43 +0000 (15:32 +0100)]
[netconfig] APT: Prefer ipv4 connections over ipv6

Ubuntu prefers ipv6 connections therefore in some networks, this
breaks software updates (it does a AAAA DNS lookup before A record
lookups). Let's prefer old style ipv4 connections over the new ipv6 in
order to save some processing and resource utilization.

Based on previous work from [1] (but without /etc/gai.conf, only APT).

[1] https://review.openstack.org/#/c/462502/

JIRA: FUEL-321

Change-Id: Ic3dff3baa1c0be9ac95972557d6a2d26641bfe1b
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
6 years ago[maas] artifact sync: improve barrier condition 51/49751/1
Alexandru Avadanii [Fri, 29 Dec 2017 01:58:25 +0000 (02:58 +0100)]
[maas] artifact sync: improve barrier condition

Simplify wait condition for MaaS service up, since it's fragile
and often adds extra time when not really needed.
Instead, retry starting boot image import right away.

Change-Id: I131d6c82127449cecf6685d4cc7484a366e658c6
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
6 years agoMerge "[patch] haproxy hostname parameterization upstream"
Alexandru Avadanii [Thu, 28 Dec 2017 15:01:31 +0000 (15:01 +0000)]
Merge "[patch] haproxy hostname parameterization upstream"

6 years ago[patch] haproxy hostname parameterization upstream 29/49729/2
Alexandru Avadanii [Thu, 28 Dec 2017 13:43:45 +0000 (14:43 +0100)]
[patch] haproxy hostname parameterization upstream

PR [1] was merged upstream.

[1] https://github.com/Mirantis/reclass-system-salt-model/pull/298

Change-Id: I335ac265b0b0b625c2f488755c5d11710ab354c2
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
6 years ago[patch] Use keystone v3 endpoints by default 19/49719/1
Michael Polenchuk [Thu, 28 Dec 2017 09:22:58 +0000 (13:22 +0400)]
[patch] Use keystone v3 endpoints by default

Change-Id: I98fc378fbec3679acf5bad4c089972340daea92c
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
6 years agoMerge "lib.sh: Add delay after `kpartx -av`"
Alexandru Avadanii [Wed, 27 Dec 2017 16:53:34 +0000 (16:53 +0000)]
Merge "lib.sh: Add delay after `kpartx -av`"

6 years agolib.sh: Add delay after `kpartx -av` 03/49703/1
Alexandru Avadanii [Wed, 27 Dec 2017 16:22:55 +0000 (17:22 +0100)]
lib.sh: Add delay after `kpartx -av`

On rare occassions, mapper bindings created by kpartx take longer
to show up, leading to errors when we try to mount them on.
Bring back the hardcoded delay to bypass such issues.

Change-Id: Ib386c04fc55cd85235a2156dba08fda378e4cdfd
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
6 years agoMerge "patches: Squash maas region changes"
Alexandru Avadanii [Wed, 27 Dec 2017 15:57:57 +0000 (15:57 +0000)]
Merge "patches: Squash maas region changes"

6 years ago[ovn] Inject ovn central options 89/49689/1
Michael Polenchuk [Wed, 27 Dec 2017 13:17:13 +0000 (17:17 +0400)]
[ovn] Inject ovn central options

Change-Id: Ib9021ee3ca15c05cc137ae42c263383acb4393bd
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
6 years agopatches: Squash maas region changes 67/49667/1
Alexandru Avadanii [Tue, 26 Dec 2017 23:35:15 +0000 (00:35 +0100)]
patches: Squash maas region changes

When re-deploying with `-f` flag, `patch -R` cannot cleanly revert
maas region changes with overlapping context lines, so squash them
into a single file.

Change-Id: I87dae72a12fea833e9e6729de21d4ce5f262695e
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
6 years ago[ovn] Mend state name of openstack 23/49623/1
Michael Polenchuk [Mon, 25 Dec 2017 09:36:51 +0000 (13:36 +0400)]
[ovn] Mend state name of openstack

* rename openstack state name with noha suffix
* increase vcpus for compute nodes

Change-Id: I03386c4c1c92d329d847aa506589823e57644ef4
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
6 years agoMerge "[ovn] Weed out gateway node from reclass storage"
Alexandru Avadanii [Fri, 22 Dec 2017 16:31:04 +0000 (16:31 +0000)]
Merge "[ovn] Weed out gateway node from reclass storage"

6 years ago[vcp] Catch 'no response' of salt minion as well 57/49557/1
Michael Polenchuk [Fri, 22 Dec 2017 13:53:36 +0000 (17:53 +0400)]
[vcp] Catch 'no response' of salt minion as well

Salt minion could return 'no response' and cause an
unconfigured state of the vcp node(s), so catch this output after linux
state as well. Also clean up excess route on proxy nodes.

Change-Id: I3183fa09ff41a8f027ee789869bdae0c3962ab8f
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
6 years ago[ovn] Weed out gateway node from reclass storage 37/49537/1
Michael Polenchuk [Fri, 22 Dec 2017 08:45:36 +0000 (12:45 +0400)]
[ovn] Weed out gateway node from reclass storage

Change-Id: I87efd87f8ac05ed9b3189e5dba80748e07c86d5d
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
6 years agoBring in ovn based scenario 35/47335/3
Michael Polenchuk [Thu, 16 Nov 2017 11:39:11 +0000 (15:39 +0400)]
Bring in ovn based scenario

OVN based scenario doesn't require conventional gateway
node since connectivity to external networks and routing
occurs on compute nodes.

Change-Id: I81e0d497170d5ffb067adf13b0e46290525f26a6
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
6 years ago[maas] Adjust deployment order/timeouts 59/49359/1
Michael Polenchuk [Wed, 20 Dec 2017 12:27:58 +0000 (16:27 +0400)]
[maas] Adjust deployment order/timeouts

Change-Id: I9dbb51ce2387450e4ae19f8b3444f5e52cfdc71d
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
6 years ago[dpdk] Remove user/group setting for ovs rundir 85/49285/3
Michael Polenchuk [Tue, 19 Dec 2017 11:35:35 +0000 (15:35 +0400)]
[dpdk] Remove user/group setting for ovs rundir

The proper patches have been merged into upstream (nova/neutron
formulas, system reclass) to use a separate dir for vhost_user sockets.

Change-Id: Iba8d8a9a05c5ab681b5b5ffbea786dca92704c82
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
6 years agoMerge "[baremetal] MaaS: Reduce timeout values"
Alexandru Avadanii [Tue, 19 Dec 2017 18:09:52 +0000 (18:09 +0000)]
Merge "[baremetal] MaaS: Reduce timeout values"

6 years agoSet libvirt unix_sock_group as an option 93/49293/1
Michael Polenchuk [Tue, 19 Dec 2017 13:59:53 +0000 (17:59 +0400)]
Set libvirt unix_sock_group as an option

Updated libvirt formula now supports group
name as an option for unix socket parameter.

Change-Id: I683e38971fe6c939fd09e95b805d611ddc596f28
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
6 years ago[baremetal] MaaS: Reduce timeout values 37/49237/3
Alexandru Avadanii [Mon, 18 Dec 2017 21:12:23 +0000 (22:12 +0100)]
[baremetal] MaaS: Reduce timeout values

`maas_fixup` is already re-entrant, so we can execute it more than
once during a commissioning/deploy cycle. Reduce the timeout waiting
for all nodes to reach a stable state, so nodes stuck in 'Ready'
state instead of reaching 'Deploying' get dealt with sooner (~5 min
vs old 30 min).

While at it, let `maas_fixup` handle machine deploy as well, so we
can catch nodes stuck in 'Ready' state and re-trigger the deploy.

Change-Id: Id24cc97b17489835c5846288639a9a6032bd320a
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
6 years agoMerge "states: networks: Use role-based addressing"
Alexandru Avadanii [Mon, 18 Dec 2017 15:39:01 +0000 (15:39 +0000)]
Merge "states: networks: Use role-based addressing"

6 years agoMerge "[baremetal] cleanup: Remove unused gtw node data"
Alexandru Avadanii [Mon, 18 Dec 2017 15:37:31 +0000 (15:37 +0000)]
Merge "[baremetal] cleanup: Remove unused gtw node data"

6 years agoMerge "[baremetal] Move salt master IP to PXE/admin"
Alexandru Avadanii [Mon, 18 Dec 2017 15:35:56 +0000 (15:35 +0000)]
Merge "[baremetal] Move salt master IP to PXE/admin"

6 years ago[maas] Set edge hwe kernel as a default minimum 99/49199/2
Michael Polenchuk [Mon, 18 Dec 2017 13:28:12 +0000 (17:28 +0400)]
[maas] Set edge hwe kernel as a default minimum

Change-Id: I360dcb675c90b6f20687979ebc493afe6682c821
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
6 years agoMerge "salt.sh: Use salt-call to apply linux sls on cfg01"
Michael Polenchuk [Mon, 18 Dec 2017 08:38:21 +0000 (08:38 +0000)]
Merge "salt.sh: Use salt-call to apply linux sls on cfg01"

6 years agoMerge "[baremetal] MaaS: Fix DHCP dynamic reserved range"
Michael Polenchuk [Mon, 18 Dec 2017 08:25:15 +0000 (08:25 +0000)]
Merge "[baremetal] MaaS: Fix DHCP dynamic reserved range"

6 years agoMerge "ci/deploy.sh: maas: cleanup_uefi on env erase"
Michael Polenchuk [Mon, 18 Dec 2017 07:23:20 +0000 (07:23 +0000)]
Merge "ci/deploy.sh: maas: cleanup_uefi on env erase"

6 years ago[baremetal] Move salt master IP to PXE/admin 55/49155/2
Alexandru Avadanii [Sat, 16 Dec 2017 22:29:54 +0000 (23:29 +0100)]
[baremetal] Move salt master IP to PXE/admin

Use PXE/admin network for salt traffic from/to all minions
except cfg01, mas01.
This allows us to drop the route to admin net from cfg01.

Change-Id: Ic2526f1ff77afe5d92ced900971f4c8f78d2d8a2
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
6 years agosalt.sh: Use salt-call to apply linux sls on cfg01 67/49167/1
Alexandru Avadanii [Mon, 18 Dec 2017 03:39:01 +0000 (04:39 +0100)]
salt.sh: Use salt-call to apply linux sls on cfg01

Also, retry applying linux high state up to twice, due to rare
spurious failures with 'No reponse' status.

Change-Id: Ic7839a5c9501673cb127412136afb91e05f87a7e
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
6 years ago[baremetal] MaaS: Fix DHCP dynamic reserved range 41/49141/3
Alexandru Avadanii [Sat, 16 Dec 2017 20:11:26 +0000 (21:11 +0100)]
[baremetal] MaaS: Fix DHCP dynamic reserved range

- patch MaaS to default to `DHCP` instead of `AUTO` for physical
  PXE interfaces (all IPs will be handed out by MaaS DHCP *inside* the
  defined dynamic DHCP IP range);
- reduce range to silence bogus MaaS warning about address exhaustion;
- regenerate pod_config.yml.example to reflect the changes;
- drop `opnfv_infra_maas_pxe_address` (duplicate of
  `opnfv_infra_maas_node01_deploy_address`);
- add `opnfv_infra_config_pxe_address` for future usage;
- while at it, fix missing patch copyright;

JIRA: FUEL-316

Change-Id: I81fad333e77f7c8508cd2b2b267c7b39c130e3e1
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
6 years ago[baremetal] cleanup: Remove unused gtw node data 39/49139/1
Alexandru Avadanii [Sat, 16 Dec 2017 18:18:45 +0000 (19:18 +0100)]
[baremetal] cleanup: Remove unused gtw node data

Change-Id: I48f73caa0f7cae532d4d9b3f68e1d8de59f8dd9e
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
6 years agostates: networks: Use role-based addressing 37/49137/1
Alexandru Avadanii [Sat, 16 Dec 2017 17:55:02 +0000 (18:55 +0100)]
states: networks: Use role-based addressing

JIRA: FUEL-310

Change-Id: Ice6d6bbb2d2ee8e21f2757b02056873d1a030ea3
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
6 years agoci/deploy.sh: maas: cleanup_uefi on env erase 07/48607/4
Alexandru Avadanii [Sat, 9 Dec 2017 01:39:33 +0000 (02:39 +0100)]
ci/deploy.sh: maas: cleanup_uefi on env erase

Running `ci/deploy.sh -EE` should also perform an UEFI boot option
cleanup, otherwise we risk booting the previously installed OS.

While at it, reduce delay between nodes removal and fix a rare failure
for `-EE` when no nodes are defined in MaaS.

Change-Id: I789ffd3e22545921216f7d5ee3509c76354542eb
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
6 years ago[baremetal] Move VCP iface config to own yml 99/48499/7
Alexandru Avadanii [Mon, 4 Dec 2017 21:33:53 +0000 (22:33 +0100)]
[baremetal] Move VCP iface config to own yml

- move linux.network.interface definitions to their own classes, which
  also removes the previous requirement of defining {dhcp,single}_nic
  parameters in classes that don't actually use them;
- drop now useless {dhcp,single}_nic parameters;
- expand linux_{dhcp,single}_interface macros, since they cause issues
  with reclass dict-merge while attempting to override sub-vars (i.e.
  it's not possible to set 'enabled: false' via reclass interpolation);

JIRA: FUEL-310

Change-Id: I29d921f545e761de335a60e242a4523d13b06c78
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
6 years agoMerge "patches: Drop salt ver sync patch, now upstream"
Alexandru Avadanii [Thu, 14 Dec 2017 17:17:37 +0000 (17:17 +0000)]
Merge "patches: Drop salt ver sync patch, now upstream"

6 years agoMerge "[baremetal] idf support for different board types"
Alexandru Avadanii [Thu, 14 Dec 2017 17:10:27 +0000 (17:10 +0000)]
Merge "[baremetal] idf support for different board types"

6 years agoMerge "Add tutorial for reclass-doc tool in documentation"
Alexandru Avadanii [Thu, 14 Dec 2017 16:51:48 +0000 (16:51 +0000)]
Merge "Add tutorial for reclass-doc tool in documentation"

6 years agopatches: Drop salt ver sync patch, now upstream 83/48983/1
Alexandru Avadanii [Thu, 14 Dec 2017 16:47:55 +0000 (17:47 +0100)]
patches: Drop salt ver sync patch, now upstream

Upstream pull request [1] was merged, drop out patch.

[1] https://github.com/salt-formulas/salt-formula-salt/pull/64

Change-Id: I7aac66349a4bec718f7821b75dbad711cd6acd58
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
6 years agoMerge "Update release notes for 5.1"
Alexandru Avadanii [Thu, 14 Dec 2017 16:23:03 +0000 (16:23 +0000)]
Merge "Update release notes for 5.1"

6 years agoReturn back repo with glusterfs 51/48951/1
Michael Polenchuk [Sat, 25 Nov 2017 16:41:20 +0000 (17:41 +0100)]
Return back repo with glusterfs

Glusterfs 3.11 package from UCA repo has the bug#875651, which
causes package instalation error due to timeout of glustereventsd
service startup. Therefore get the latest version from PPA repo.

Change-Id: Ia47d969e758c8689508dfd6b7fd2725514ea88c5
Co-Authored-By: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
6 years ago[baremetal] idf support for different board types 45/48945/2
Guillermo Herrero [Thu, 14 Dec 2017 10:36:35 +0000 (11:36 +0100)]
[baremetal] idf support for different board types

This modification enables having different models of target nodes
but restricts them to two groups of the same kind assuming:
 - 3 KVMs will be same model and have the same NIC names
 - 2 Compute nodes will be the same model and have same NIC names

This is an ad-hoc change for arm-pod7. Will work for all other setups
but the jinja2 logic should be evetually extended to support a per
board approach, allowing any board to be any kind

Change-Id: Ie994d661e3dbe5b9a33d290a850a6152cf320c86
Signed-off-by: Guillermo Herrero <guillermo.herrero@enea.com>
6 years agoMerge "[baremetal] Upgrade packages on kvm, cmp nodes too"
Michael Polenchuk [Thu, 14 Dec 2017 08:07:30 +0000 (08:07 +0000)]
Merge "[baremetal] Upgrade packages on kvm, cmp nodes too"

6 years agoRevert "Horizon: service: Default to v2 API" 81/48881/1
Alexandru Avadanii [Thu, 14 Dec 2017 02:05:20 +0000 (03:05 +0100)]
Revert "Horizon: service: Default to v2 API"

Pike Horizon packages contain the fix for full v3 compatibility.
Also update the reclass models to use v3 API.

JIRA: FUEL-284

This reverts commit 68ff467c4c2d9d0f201e2912c97616402f8109a3.

Change-Id: I7795197100a67f892889eefce73252ec90d289cc
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
6 years ago[baremetal] Upgrade packages on kvm, cmp nodes too 75/48875/1
Alexandru Avadanii [Thu, 14 Dec 2017 01:33:28 +0000 (02:33 +0100)]
[baremetal] Upgrade packages on kvm, cmp nodes too

MaaS installs a few packages via curtin, which don't get upgraded
later, even if we add extra repos (it will be the case for novcp).
Perform a package upgrade on all baremetal nodes, as it's also a
good security practice.

JIRA: FUEL-310

Change-Id: Ib365e42a0d2f65813fc9bef516326ec89fe407de
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
6 years agodeploy.sh: Fix missing space in dependencies list 51/48851/1
Alexandru Avadanii [Wed, 13 Dec 2017 19:59:46 +0000 (20:59 +0100)]
deploy.sh: Fix missing space in dependencies list

Change-Id: I4d41ff8b9f7c1b59a2c1a400b9335f17b8c542be
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
6 years agoMerge "deploy.sh: Factor out pkg deps into yaml files"
Alexandru Avadanii [Wed, 13 Dec 2017 19:29:02 +0000 (19:29 +0000)]
Merge "deploy.sh: Factor out pkg deps into yaml files"

6 years agopatches: libvirt: Use "libvirt" unix_sock_group 47/48847/2
Alexandru Avadanii [Sun, 20 Aug 2017 16:18:53 +0000 (18:18 +0200)]
patches: libvirt: Use "libvirt" unix_sock_group

Now that libvirt 3.6 is used on kvm nodes (via UCA repos), which
by default uses "libvirt" group for socket ownership,
change old "libvirtd" in salt-formula-libvirt's:
- libvirt/files/libvirtd.conf.Debian

This allows us to remove the manual group creation from VCP state
file, which was not re-entrant (failed if group already existed).

Change-Id: Id61fecd82daec1c0716ff4796b79dce47d096c3a
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
6 years ago[baremetal] Link br-ex as ovs port 99/48799/2
Michael Polenchuk [Wed, 13 Dec 2017 12:00:29 +0000 (16:00 +0400)]
[baremetal] Link br-ex as ovs port

Add br-ex and physical public interface
as ports directly to ovs bridge.

Change-Id: I12c078d3e286af7c5fa3078cc055905c03cf600b
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
6 years agoAdd tutorial for reclass-doc tool in documentation 17/48517/8
Charalampos Kominos [Thu, 7 Dec 2017 10:19:04 +0000 (11:19 +0100)]
Add tutorial for reclass-doc tool in documentation

JIRA: FUEL-293

Change-Id: I2612e8d63875dca4d05f8c30db24a5ae84cc8dea
Signed-off-by: Charalampos Kominos <Charalampos.Kominos@enea.com>
Signed-off-by: Cristina Pauna <cristina.pauna@enea.com>
6 years agoUpdate release notes for 5.1 03/48803/1
Cristina Pauna [Wed, 13 Dec 2017 15:17:49 +0000 (17:17 +0200)]
Update release notes for 5.1

The release notes are updated with relevant Jira filters and
changes since 5.0

Change-Id: If14cb0369ced899d17608ef43e93b848e8b2a5c8
Signed-off-by: Cristina Pauna <cristina.pauna@enea.com>
6 years agoAArch64: Move to pike-armband repo 57/48757/2
Alexandru Avadanii [Tue, 12 Dec 2017 22:07:56 +0000 (23:07 +0100)]
AArch64: Move to pike-armband repo

Armband restructured its repos to support UCA-only virtual deploys,
propagate the change to our base image pre-build. Drop apt-mk repos,
should only be added later by Salt, when MCP repos are also present.

JIRA: ARMBAND-338

Change-Id: Ieafa6d3fb0599679341053187b4a7d778af86428
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
6 years agodeploy.sh: Factor out pkg deps into yaml files 51/48751/6
Alexandru Avadanii [Sat, 9 Dec 2017 17:20:16 +0000 (18:20 +0100)]
deploy.sh: Factor out pkg deps into yaml files

Instead of hardcoding the packages to install on the jumphost at
deploy time in `ci/deploy.sh`, move the requirements list to a
dedicated yaml file, so it is decoupled from actual code.

While at it, add missing deps:
- ipxe-qemu (AArch64 Ubuntu only);

Change-Id: I9ee5c0796b70f9d29427d608b6415333c8841e1e
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
6 years agoMerge "[baremetal] Move all MaaS PXE net config to PDF"
Alexandru Avadanii [Tue, 12 Dec 2017 15:51:22 +0000 (15:51 +0000)]
Merge "[baremetal] Move all MaaS PXE net config to PDF"

6 years agoMerge "[baremetal] Fixup pike deployment (2)"
Alexandru Avadanii [Tue, 12 Dec 2017 15:42:14 +0000 (15:42 +0000)]
Merge "[baremetal] Fixup pike deployment (2)"

6 years ago[baremetal] Fixup pike deployment (2) 01/48701/1
Michael Polenchuk [Tue, 12 Dec 2017 09:00:30 +0000 (13:00 +0400)]
[baremetal] Fixup pike deployment (2)

* align dpdk packages list
* add missing libvirtd group to vcp host nodes

Change-Id: Ideab2b16ca0561035b225e58bf3d0c5653d303bb
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
6 years ago[baremetal] Set public vlan tag on ovs port 97/48697/1
Michael Polenchuk [Tue, 12 Dec 2017 08:48:28 +0000 (12:48 +0400)]
[baremetal] Set public vlan tag on ovs port

Change-Id: Ia5a34d8d8a9649fc911198fed86aee51071ba568
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
6 years agoMerge "AArch64: s/ocata/pike/g in scenario defaults"
Alexandru Avadanii [Mon, 11 Dec 2017 19:40:11 +0000 (19:40 +0000)]
Merge "AArch64: s/ocata/pike/g in scenario defaults"

6 years agoAArch64: s/ocata/pike/g in scenario defaults 69/48669/1
Alexandru Avadanii [Mon, 11 Dec 2017 18:56:54 +0000 (19:56 +0100)]
AArch64: s/ocata/pike/g in scenario defaults

We hard set some repo URLs used for base image pre-building, so bump
those to Pike, now that Armband repos for Pike are populated.

JIRA: ARMBAND-335

Change-Id: Ibd7267b628985a3c7a33d9d8a93c824077c69f10
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
6 years ago[baremetal] Move all MaaS PXE net config to PDF 75/48575/5
Alexandru Avadanii [Fri, 8 Dec 2017 00:20:47 +0000 (01:20 +0100)]
[baremetal] Move all MaaS PXE net config to PDF

- s/opnfv_maas_pxe_/opnfv_infra_maas_pxe_/g to align with other vars;
- patches: pharos: Add MaaS PXE network to installer adapter;
- runtime.yml{,.template}: move to installer adapter, update
  pod_config.yml example;
- drop MAAS_PXE_NETWORK global env var, now read strictly from PDF;

JIRA: FUEL-313

Change-Id: I46d7510bd53fba7890c411d36bc28fd6ff6f3648
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
6 years agopatches: pharos: Bump & drop upstream patches 65/48665/2
Alexandru Avadanii [Mon, 11 Dec 2017 18:15:15 +0000 (19:15 +0100)]
patches: pharos: Bump & drop upstream patches

All our staged patches for Pharos repo are now upstream, so bump
the git submodule and remove the patch files.

Change-Id: I0d68eb3bdd9abfa286c3640acc1f13ce6100801d
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
6 years agoMerge "[baremetal] cleanup: Parameterize VCP hostnames"
Alexandru Avadanii [Mon, 11 Dec 2017 17:41:37 +0000 (17:41 +0000)]
Merge "[baremetal] cleanup: Parameterize VCP hostnames"

6 years agoMerge "[baremetal] Fixup pike deployment"
Alexandru Avadanii [Mon, 11 Dec 2017 17:15:05 +0000 (17:15 +0000)]
Merge "[baremetal] Fixup pike deployment"

6 years agoMerge "pod_config: cleanup: Drop infra_compute_*_address"
Alexandru Avadanii [Mon, 11 Dec 2017 17:11:48 +0000 (17:11 +0000)]
Merge "pod_config: cleanup: Drop infra_compute_*_address"

6 years ago[baremetal] cleanup: Parameterize VCP hostnames 19/48619/3
Alexandru Avadanii [Sun, 10 Dec 2017 19:24:01 +0000 (20:24 +0100)]
[baremetal] cleanup: Parameterize VCP hostnames

We already define the hostnames in a central location
(openstack_init), so use the reclass params instead of hardcodes.
Also handle hardcoded hostnames in system.haproxy.proxy.listen
and system.mongodb.server.cluster.

This will also simplify later changes for novcp scenarios.

JIRA: FUEL-310

Change-Id: I2a8b4cd1fcb61a7c63f9047123d985bb688dca0a
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
6 years agoRevert "Bring in newer glusterfs for mtime unsplit brain" 53/48653/2
Michael Polenchuk [Mon, 11 Dec 2017 15:13:25 +0000 (15:13 +0000)]
Revert "Bring in newer glusterfs for mtime unsplit brain"

Since the uca/pike repo has version of 3.11
This reverts commit 8a867449217151f59766a74c6165ffc26c7f50b4.

Change-Id: I5cc0d94b4a1b53b987fab03bc20466ee3c3eb8b2
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
6 years ago[baremetal] Fixup pike deployment 59/48659/1
Michael Polenchuk [Mon, 11 Dec 2017 15:30:17 +0000 (19:30 +0400)]
[baremetal] Fixup pike deployment

* up glusterfs.client state before actual volume usage
* handle keystone server state
* specify suitable nova packages list for uca repo
  (consoleproxy vs novncproxy)
* upgrade vcp nodes to get proper cryptography
  library for keystone
* align service names for libvirt & glusterfs

Change-Id: Iaeb7d147e6d407bbeaec2d40fd81037c939c3fe0
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
6 years agopod_config: cleanup: Drop infra_compute_*_address 15/48615/2
Alexandru Avadanii [Sun, 10 Dec 2017 18:37:13 +0000 (19:37 +0100)]
pod_config: cleanup: Drop infra_compute_*_address

Our PDF installer adapter defines 3 unused reclass params that are
misleading, so drop them (infra_compute_node{01,02,03}_address).

Change-Id: I0ca49b73add543c8f03b14ed4d805ce55f973bb5
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
6 years agoMerge "[baremetal] Update openstack version to pike"
Michael Polenchuk [Mon, 11 Dec 2017 09:59:24 +0000 (09:59 +0000)]
Merge "[baremetal] Update openstack version to pike"

6 years agoMerge "states: maas: Stop using maas-stable PPA"
Michael Polenchuk [Mon, 11 Dec 2017 09:23:46 +0000 (09:23 +0000)]
Merge "states: maas: Stop using maas-stable PPA"

6 years ago[baremetal] Update openstack version to pike 87/48387/2
Michael Polenchuk [Tue, 5 Dec 2017 09:40:58 +0000 (13:40 +0400)]
[baremetal] Update openstack version to pike

Change-Id: If4208e937734df72b10aad259c65a2617ff5e19c
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
6 years agoMerge "[baremetal] Migrate public bridge to ovs"
Michael Polenchuk [Mon, 11 Dec 2017 08:16:58 +0000 (08:16 +0000)]
Merge "[baremetal] Migrate public bridge to ovs"

6 years agostates: maas: Stop using maas-stable PPA 13/48613/2
Alexandru Avadanii [Sat, 9 Dec 2017 20:40:22 +0000 (21:40 +0100)]
states: maas: Stop using maas-stable PPA

Currently, Xenial repos provide MaaS 2.2.x, while the PPA bumped it
to 2.3.x. Since we switched to 2.3, we observed a rare wrongful state
transition from 'Deploying' back to 'Ready'.
Drop the PPA, falling back to 2.2 from mainline distro repos.

JIRA: FUEL-312

Change-Id: I3daa118059f37cbeca076da685661c28f3a28a97
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
6 years agosubmodule: pharos: Bump to fix run without eyaml 05/48605/2
Alexandru Avadanii [Sat, 9 Dec 2017 00:21:42 +0000 (01:21 +0100)]
submodule: pharos: Bump to fix run without eyaml

Most Fuel PODs do not use encrypted PDFs, so bump Pharos git
submodule to include updated `generate_config.py` to silently
skip decryption in case eyaml is not installed.

Change-Id: I845a32c7001123a93e38197e0daf53d18c797aed
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
6 years ago[baremetal] Migrate public bridge to ovs 95/48595/1
Michael Polenchuk [Fri, 8 Dec 2017 14:06:57 +0000 (18:06 +0400)]
[baremetal] Migrate public bridge to ovs

Get rid of "ovs-to-linux" links, instead of this setup native ovs
bridge for public interface and link it with neutron external bridge.

Change-Id: I483881e78f22fc035107a622307d95761c4a5667
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
6 years ago[baremetal] PDF: Use net_admin for MaaS PXE net 73/48573/4
Alexandru Avadanii [Thu, 7 Dec 2017 22:24:37 +0000 (23:24 +0100)]
[baremetal] PDF: Use net_admin for MaaS PXE net

While at it, fix another small issue where a variable
(parameters__param_opnfv_maas_pxe_address) was used without being
populated (defined in 'runtime.yml', which is not read using
`parse_yaml`, like the scenario yml or the PDF expanded class,
pod_config.yml).

JIRA: FUEL-313

Change-Id: Iee88601d5420f55572bf90c8cf330afbedfb2e21
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
6 years agopatches: pharos: Drop patches merged upstream 65/48565/1
Alexandru Avadanii [Thu, 7 Dec 2017 20:02:02 +0000 (21:02 +0100)]
patches: pharos: Drop patches merged upstream

Also, remove redundant pharos patch that adds prx mgmt IPs:
- "Re-assign mgmt network to proxy nodes"
as those values are set already (to different values!) by patch:
+ "extend public gateway support"

While at it, `make patches-export` should clean the patch dir first.

Change-Id: Ice106e5d48c7b4cd90ffc6af7441199034d4f546
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
6 years agoMerge "virtual: Add infra/init.yml for each scenario"
Alexandru Avadanii [Thu, 7 Dec 2017 14:44:39 +0000 (14:44 +0000)]
Merge "virtual: Add infra/init.yml for each scenario"

6 years agoMerge "base image: Explicitly call kpartx"
Michael Polenchuk [Thu, 7 Dec 2017 08:15:12 +0000 (08:15 +0000)]
Merge "base image: Explicitly call kpartx"

6 years agoMerge "PDF: reclass: Rename template file suffix to _pdf"
Michael Polenchuk [Thu, 7 Dec 2017 08:15:00 +0000 (08:15 +0000)]
Merge "PDF: reclass: Rename template file suffix to _pdf"

6 years agoMerge "lib.sh: persistent cfg01, mas01 mcpcontrol"
Michael Polenchuk [Thu, 7 Dec 2017 07:56:28 +0000 (07:56 +0000)]
Merge "lib.sh: persistent cfg01, mas01 mcpcontrol"

6 years agoClean up noifupdown support patch 79/48379/1
Michael Polenchuk [Tue, 5 Dec 2017 08:44:50 +0000 (12:44 +0400)]
Clean up noifupdown support patch

The patch has been merged to upstream linux formula.

Change-Id: I09f96be920c5c3ecd09fd2ab132e25519902239b
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
6 years agovirtual: Add infra/init.yml for each scenario 21/48321/1
Alexandru Avadanii [Sat, 2 Dec 2017 23:39:05 +0000 (00:39 +0100)]
virtual: Add infra/init.yml for each scenario

Align our reclass model for virtual PODs with baremetal, by adding
an infra init file for each scenario, setting up cluster_name via
scenario infra init instead of scenario init.
While at it, reduce redundancy by defining cluster_domain based on
cluster_name via common infra init.

JIRA: FUEL-310

Change-Id: I5e89c883853fa66cb1c1fc69ce5766ee136ac477
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
6 years agoPDF: reclass: Rename template file suffix to _pdf 27/48027/2
Alexandru Avadanii [Wed, 29 Nov 2017 03:47:48 +0000 (04:47 +0100)]
PDF: reclass: Rename template file suffix to _pdf

Init is a generic suffix which also collides with pre-existing files.
A distinct suffix (e.g. _pdf) should be used.

JIRA: FUEL-310

Change-Id: I919a5394e7ff432ee86d10bafe889aeadb181649
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
6 years agostates: Rename openstack, add baremetal_init 17/48317/1
Alexandru Avadanii [Fri, 1 Dec 2017 02:03:52 +0000 (03:03 +0100)]
states: Rename openstack, add baremetal_init

To align with new cluster naming convention, rename 'openstack'
state file to 'openstack_noha'.
While at it, factor out baremetal setup from 'virtual_control_plane'
into a new state that will be reused in upcoming scenarios, remove
useless sync_all (automatically done after node reboot).

FUEL-310

Change-Id: I6d7e5db8f09305f2fd8eeca0199a2e85b08d2202
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
6 years agobaremetal: openstack_proxy: Rename iface dicts 15/48315/1
Alexandru Avadanii [Sun, 3 Dec 2017 18:44:32 +0000 (19:44 +0100)]
baremetal: openstack_proxy: Rename iface dicts

Later we will want to merge multiple openstack_* classes, which would
try to merge the 'single', 'dhcp' interface definitions and fail due
to mismstached types (proxy has 'dict' types while the rest do not).
Let's rename the proxy iface definitions to prevent that.

JIRA: FUEL-310

Change-Id: I1be18ddbbeae1bb3ee6de9bd783ee57b185cd477
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
6 years agobaremetal: Use common database_init, control_init 13/48313/1
Alexandru Avadanii [Sun, 3 Dec 2017 01:58:11 +0000 (02:58 +0100)]
baremetal: Use common database_init, control_init

While at it, delete unused haproxy_openstack_api.

JIRA: FUEL-310

Change-Id: I01953b01624fe3e9da8074239d20056f3762efc7
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
6 years agoRename reclass models to add "-ha" or "-noha" 81/47981/4
Alexandru Avadanii [Tue, 28 Nov 2017 19:27:05 +0000 (20:27 +0100)]
Rename reclass models to add "-ha" or "-noha"

Parse all reclass j2 templates, not only common + current scenario
(useful when adding new scenarios later).

JIRA: FUEL-310

Change-Id: I8e87af702f83c42cb8f766bf6f121449aa5f2c26
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
6 years agoMerge "[virtual] Update dpdk scenario with pike"
Michael Polenchuk [Mon, 4 Dec 2017 09:46:19 +0000 (09:46 +0000)]
Merge "[virtual] Update dpdk scenario with pike"