[AArch64] Pin armband repo with highest prio

Mirantis repositories shadow certain packages from Armband repos
(e.g. qemu-efi) if pinning is not set.
Since MaaS does not allow configuring repo priorities via its API,
use Salt pillar data to re-define the Armband repo and set its prio.

Change-Id: Ic743f4b684b3f13552792f5f04097fac73171b37
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>

Revert "[salt-formulas] Switch to 'testing' dist"

This reverts commit a8f434118013a69295bdc0f9fc8294c539742c01.
This reverts commit b52ef6ea62f43de4b91e2becd92a3282e569925c.

Change-Id: I604df3f50febb049f1545a8141c5abb013cf6c69
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>

Merge "[patch] MaaS: fabric ID from CIDR"

[patch] MaaS: fabric ID from CIDR

Change-Id: I1e1490f4d0a0bee2783450c6369d7ab6c45b193f
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>

sysinfo_print: Dump all iptables rules

Change-Id: I0ce6d21e14efe6e5236759c8d8b3439f09e89675
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>

[salt-formulas] Also switch reclass to 'testing'

Previous commit only changed the dist during Salt master bootstrap,
but failed to update the DEB repo configration in reclass.

Fixes: a8f43411

Change-Id: Ie011719cdc1c75cba363f53e51f22ce7cf138a5e
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>

[salt-formulas] Switch to 'testing' dist

Nightly repo dist is currently broken.

Change-Id: I9235c2c95b4edebf56d20e33932237f18a8e63d5
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>

[submodule] Bump Pharos for arm-virtual2 ip-range

Change-Id: I77859a2670e07668913232d482853d77cfc8b149
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>

Update release notes for 6.2

Change-Id: Ibd168ef31af1f8a9a2c128ce8a8e95a5c231a917
Signed-off-by: Cristina Pauna <cristina.pauna@enea.com>

Merge "[noha] Adopt vPDF usage"

Merge "[ha] Add fabric to maas admin subnet"

[ha] Add fabric to maas admin subnet

Change-Id: Ida4cf736dfe60886384e7e180d3e43faf811bd58
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>

[ha] Pin UCA's libvirt/qemu on computes

* install libvirt/qemu packages only from UCA on computes
* shift panko/aodh db settings to telemetry nodes

Change-Id: I0f4434aa8d2139870afaed9d9831db0a9af10e51
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>

[noha] Adopt vPDF usage

Previously, all virtual deployments relied on lab configuration from
<./mcp/config> using 'local-virtual1' as POD name.
To keep using that default sample config, one can still pass the
appropiate arguments to <ci/deploy.sh> as in:
$ ./ci/deploy.sh -l local -p virtual1 [...]

This changed introduces support for vPOD-specific lab config, e.g.:
$ ./ci/deploy.sh -l ericsson -p virtual4

Change-Id: I6ecb8e71d7e923e4b29b4ab0693c1afa5e19b582
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>

[MaaS] Adopt boot-resources control from maasng

JIRA: FUEL-364

Change-Id: I891514f85cf694509003a3b0a6f3568524d0a461
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>

[ha] Update pymysql to version 0.8

* get latest pymysql from UCA repo
* set DB connection recycle time to avoid issue with broken database
  connection during query (MySQL server has gone away)
* set haproxy source balacing mode for neutron API to avoid race
  condition with duplicate security groups
* get libvirt/qemu from UCA repo on compute nodes since main process
  of libvirt exits (is killed by 11/SEGV [segfault]) during volume
  attaching operation

Change-Id: Iead5ecc561683876c476357f4a4de8780775aa7b
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>

[submodule] Bump Pharos for unh-pod1 updates

Change-Id: Ic6705c64dd3582b562e96ff79fd7fc83789533a6
Signed-off-by: Paul Vaduva <Paul.Vaduva@enea.com>
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>

[ha] cmp: Fix duplicate PXE/admin IDs (2)

Certain scenarios build additional interface definitions which
overlap with the ones from common-ha reclass dir.
Extend previous fix to scenario-specific interface definitions
to prevent such duplicates.

Change-Id: I57b175e2f760a22454723266b044845335af4cfe
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>

[MaaS] curtin: Switch back to Salt 2016.11 repo

Change-Id: I7482a3b7045e6e1e94d6abfe2dc6b862f2176a53
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>

[noha] Rectify opendaylight+dpdk scenario

Change-Id: Ie707de90617e7080a5eb87cad69a604cf5e0f9fc
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>

[ha] cmp: Fix duplicate PXE/admin interface IDs

Change-Id: I99a8acfbfdddb79493bf3a191835927cef5fd5e3
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>

[queens] Update Gnocchi version

Change-Id: Ieea2c510ae4c6d4db0667f3341c877042a2c1846
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>

Enforce static configuration instead of DHCP

- noha: 'accept_policy: open_mode' to align with ha scenarios;
- s/cmp01/cmp001/g to align all scenarios and allow code reuse;
- rename network params: s/dhcp/mcpcontrol/g, cleanup;
- computes XDF data: drop 'opnfv_*' layer of params, cleanup;
- local vPDF: add comments with default roles by node index;
- parameterize all netmasks;
- drop unused address/netmask for 'proto: manual' interfaces;
- virsh_net: cleanup definitions, remove hardcodes, align IP on
  jumpserver and DHCP range with MaaS for pxebr;
- maas: parameterize hardcoded '/24' cidr for PXE/admin, refactor
  maas.region.machines parameterization;
- merge <all-mcp-arch-common/infra/config_*pdf.yaml.j2> templates;
- move reclass.storage definitions of compute nodes to common dir;
- drop 'openstack_compute_*' reclass params in favor of expanding
  them via j2 directly in reclass.storage params;
- adopt `nm.cluster.has_*_nodes` where possible;
- obsolete `runtime.yml` from reclass model;
- refactor arch-specific reclass param selection;
- remove unused defaults in favor of mandatory IDF properties;
- noha: prepare for baremetal node support in cinder_lvm_devices;
- interfaces: add interface_mtu and 'noifupdown: true' everywhere;
- interfaces: use j2 macros to generate eth/vlan config;
- states cleanup: remove DHCP route disable workaround on prx/cmp;
- allow configuring NTP servers via:
  `idf.fuel.network.ntp_strata_host{1,2}`;
- ovs_bridge: Allow setting gateway, dns-nameservers
- apache: Adjust module list for novcp class inheritance;
- glusterfs PPA: pin with same prio of MCP repos for novcp scenario;

JIRA: FUEL-319
JIRA: FUEL-326
JIRA: FUEL-337

Change-Id: Ia6ad64ba8cade85a75fb22c9a2505decc3834360
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>

Merge "[patch] OVS bridges should not use 'auto'"

Rectify nova placement API check

* make a GET request to placement service as it returns 200 with
  version data starting from nova 17.0
* turn on nova image preallocation mode on baremetal computes
* double salt worker threads by number of available CPU

Change-Id: I509cbab3b4b8460452f5e487e457edfedd5da29a
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>

[patch] OVS bridges should not use 'auto'

Change-Id: I0db7b9ee5af77f9e2d580c4ace9fa7d7f17c8d85
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>

Get back to salt 2016.11

Change-Id: Id024ed22dd1760f41ae18aeb8e680c2f07a5dc63
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>

[patch] Rebase keyserver proxy support (3)

Rebased on top of [1].

[1] https://github.com/salt-formulas/salt-formula-linux/commit/9f30456

Change-Id: Iad5aef674c47d622a94d1c21cae3f46fbb3c52d8
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>

Merge "[ha] Fix salt state apply on ctl VCP nodes"

[ha] Fix salt state apply on ctl VCP nodes

Sphinx metadata collection does not check whether a haproxy server
is enabled or not, so keep full 'heat_cloudwatch_api' configuration
in place to prevent breakage; and just disable it instead.

Change-Id: I9fda1f3d2829b89ea8e425ae831a54484651f60a
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>

[ha] Retry salt.minion state apply

Baremental compute nodes time out with 'Not connected' or
'No response' while trying to apply `salt.minion` highstate after
the Salt 2017.7 bump, retry applying it up to 5 times.

Change-Id: I67f74c65a7cebe0a0b4c2812fa0ced6c1144be58
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>

Merge "Smooth down API proxy config"

Merge "[state] Stop expecting linux.network failures"

Smooth down API proxy config

* heat: remove deprecated AWS compatible CloudWatch API
* nova: placement API responds with 405 when doing GET to root URL

Change-Id: I545e5b0fc5b57b516a4b1d8c9e33ea1d5d61ed76
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>

[patch] Rebase keyserver proxy support

Rebase on top of [1].

[1] https://github.com/salt-formulas/salt-formula-linux/commit/6f5e69e

Change-Id: Ic34bb9c1fad8778262f2e876c91667e903d143bf
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>

[state] Stop expecting linux.network failures

Now that the `linux.network` state applies cleanly on all nodes,
stop expecting failures to prevent false-positive passes.

Change-Id: I8a283474e12b85deb95815e4e7a0af4a74d143ac
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>

Merge "Bring in Tacker salt formula"

Merge "Discover computes after they are registered"

Bring in Tacker salt formula

Install tacker from git repository since there is no ubuntu package yet.

Change-Id: Ibe4b6486050213df1a545c5c79c43a635bbf6c08
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>

Unmask OpenDaylight service explicitly

After salt update to version 2017.7.0 the indefinite mask
has to be removed before attempting to start the service.

Change-Id: I21616929f06f8ebd8a2d70e8c33f92c7b808a9c5
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>

Discover computes after they are registered

The "nova hypervisor-list" won't work (i.e. will result in an empty list)
until discover_hosts is launched with compute nodes on board.

Change-Id: I942c69645731b59a5cb9e5cf39a89b84def96b12
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>

[ha] Update OpenStack version to Queens

* make cluster names version-agnostic
* switch to mirror.mirantis.com
* system reclass update

Change-Id: I96394b3c9d8d38321289af8e3a4e8bbc5dd7a408
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>

[noha] Update OpenStack version to Queens

* make cluster names version-agnostic
* pre-install oslo-templates formula
* salt version update to 2017.7
* system reclass update

Change-Id: I903895f6bcd98a1062e66e3f14d23ec04dc1b869
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>

[patch] Rebase APT keyserver proxy support patch

Sync our patch with upstream PR [1].

[1] https://github.com/salt-formulas/salt-formula-linux/pull/138

Change-Id: If3a51d3aef51accee0901c59e1342b42dbe4be4b
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>

Clean out merged upstream patch

Change-Id: I0c8f87bad0fbe55684bd43547674ed91a31f39f8
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>

[submodule] Bump Pharos for ericsson-pod1 updates

This bump also brings in:
- arm-pod{8,9} config;
- UNH lab config;
- NTP config in IDFs for existing PODs;
- basic vPDF support;

Change-Id: Ifc16616bc8a13de2bdf048971448d4853e93880e
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>

[patch] Drop upstream reclass formula patches

Change-Id: Iad2a27d059b43ed14fb70bdee01b3db29613615b
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>

[patch] maas: rebase artifact download

Upstream refactored maas/regions.sls to introduce kitchen tests in
[1], rebase our patches on top of it.

[1] https://github.com/salt-formulas/salt-formula-maas/commit/8a0d52e

Change-Id: I491fb2e05679ebc226a27141e685d429e0ff8bcc
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>

[noha] Disable proxy certificate generation

noha scenarios try to generate the proxy SSL cert using the Salt
Master CA authority. However, this requires a `salt-minion` restart
between the CA configuration and the cert generation (see [1]).
Since we don't use this cert anyway, let's just disable it for cfg01.

JIRA: FUEL-377

[1] https://github.com/salt-formulas/salt-formula-salt/issues/67

Change-Id: I8e88200e0d5614f0034c0c77e6ba675dc7ef0c11
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>

mcp: salt-formulas: opendaylight: AArch64 leveldb

ODL requires native leveldbjni support on architectures like AArch64,
provided as a Debian package in ODL Team Nitrogen PPA.

Only systemd is supported (unlikely to change).

JIRA: ARMBAND-387

Change-Id: Ie7f2955c6574ab4584ed0c207b42ed7ab7261561
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>

[maas] Download only used subarches images

- reduce download size/time for MaaS images by downloading only the
  subarch images we use;
- fix stop condition in MaaS artifact download script;
- stop importing images before updating the boot resources to prevent
  race conditions;

Change-Id: I3b15fae3463bc6c54be2cc06e0cb31edcbd307cb
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>

Merge "[Horizon] Drop the obsolete Horizon workaround"

Merge "AArch64: Switch back to hwe-16.04 kernel"

Merge "[maas] Adopt maas.machines.storage"

maas: boot-resources: Dynamic arch list from PDF

Instead of leaving MaaS default to 'amd64', explicitly set the list
of architectures to support based on PDF data.

Change-Id: I852a3ce156db3df0c090f10c0b45c26058dbb6c6
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>

[maas] Adopt maas.machines.storage

Replace MAAS CLI set_disk_layout with the
new maas.machines.storage state

JIRA: FUEL-364

Change-Id: I4d8cd9f473c5386ee7b32ad378ca1e02989233ca
Signed-off-by: ting wu <ting.wu@enea.com>

AArch64: Switch back to hwe-16.04 kernel

Align kernel versions across architectures (where possible).

Change-Id: I66a822611eb5e46f90b62f5d36df571ae75dcba3
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>

Merge "[docs] Update documentation for Fraser-6.1"

[docs] Update documentation for Fraser-6.1

Change-Id: Ic5e71c0ce2bcfd9d01f07fff017aac8019793527
Signed-off-by: Charalampos Kominos <Charalampos.Kominos@enea.com>

[Horizon] Drop the obsolete Horizon workaround

Since we switch the Horizon package from
Ubuntu/UCA repo to MCP repos,  we can drop the
obsolete Horizon workaround.

JIRA: FUEL-374

Change-Id: I0b77a4f2158a972a5eb400919f2929c7bd8360cf
Signed-off-by: ting wu <ting.wu@enea.com>

Revert "[patch] Backport proposed generate_dhparams fix"

This reverts commit c7dff6e1ae6444f60d3fe20b1f495609c7980051.
The patch has been merged into upstream.

Change-Id: Idf768b9b94b3498994601d50feeeef008430ba27
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>

[patch] Backport proposed generate_dhparams fix

Upstream PR [1] should fix recently introduced issues with nginx
state for SSL-enabled sites.

[1] https://github.com/salt-formulas/salt-formula-nginx/pull/40

Change-Id: I52b5e4f51539e535249e0850f0b34c2801f4d74a
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>

Merge "[maas] AArch64: Preseed Armband repositories"

[maas] AArch64: Preseed Armband repositories

Preseed Armband common repo-comp for Pike, so we get
the updated kernel & other packages from the start.
Requires upstream PR [1].

While at it, also handle related PR merged upstream [2] by
explicitly setting maas.region.subnets name via j2 instead of reclass
param expansion in name (allows us to drop the remaining chunk of the
fabric-from-CIDR patch we used to carry).

Note: opnfv_infra_maas_pxe_network_addres can now be dropped from
pod_config j2 template in Pharos Fuel installer adapter.

[1] https://github.com/salt-formulas/salt-formula-maas/pull/26
[2] https://github.com/salt-formulas/salt-formula-maas/pull/22

Change-Id: I356adb73b80f0f2d85db4ab060e804cb9a053862
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>

[docs] Update guest OS support matrix

Re-testing the supported operating systems for the guests
revealed that Fedora is now working without problems or user input
on aarch64.

Change-Id: I87dcac2d3297b7a43dd252d4aecfdea4254e519b
Signed-off-by: Charalampos Kominos <Charalampos.Kominos@enea.com>

[ha] Bring in gnocchi/panko services

JIRA: FUEL-372
Change-Id: I9083898f0179649d855750d48f3f9bc79caba50c
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>

[ha] Align package repositories on all the nodes

Change-Id: Idd09d28677b109b4c0c5a4c232dcfb7c4f24dbd7
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>

Merge "[noha] Bring in gnocchi/panko services"

Merge "[docs] add info about block storage support"

[noha] Bring in gnocchi/panko services

JIRA: FUEL-372
Change-Id: I4e322a4a2c84843e9350fe9b3b849cd0c5244a12
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>

[noha] Align package repositories on all the nodes

* replace UCA repo with MCP one on the remaining nodes
* clean up haproxy out-of-use frontends:
  - Nova EC2 API
  - Glare API

Change-Id: If8e6de2d95387c92ae96ad6069ec788ab229e1ce
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>

Merge "AArch64: Switch to upstream Saltstack repo"

AArch64: Switch to upstream Saltstack repo

To work around the missing dependency of 'python-tornado' on
'python-features' provided by the binary-specific DEB package
in amd64 Saltstack repository, pre-install the req explicitly
on FN (jumpserver) and VCP VMs via lib.sh.

JIRA: ARMBAND-382

Change-Id: I2dbd08a58f6cf57c2c48c2d7fc8d19ad4872e800
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>

[docs] add info about block storage support

Adding volumes in openstack by default uses virtio_blk which limits
the amount of disks that can be attached. Here we inform about
this shortcoming and supply a workaround.

JIRA: ARMBAND-376

Change-Id: I711df5c19befca7a3abfdce9fe9b2b0bb38853ac
Signed-off-by: Charalampos Kominos <Charalampos.Kominos@enea.com>

[xdf] Fix parsing of non-range private vlan(s)

Private (tenant) vlan range support introduced a regression in parsing
PDF/IDF with a single vlan (no range) due to the value being
interpreted as an int instead of string.

Change-Id: Id5d67c4c516a9bb807ffa5e111ead5cbcee7e2ca
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>

Merge "Update PTL"

[virtual] Bring in OpenDaylight DPDK scenario

Change-Id: If77ac85fa86e0a1a18c0cc2abff77d876cdb9e93
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>

Update PTL

Election process
https://lists.opnfv.org/pipermail/opnfv-tech-discuss/2018-May/021163.html

Change-Id: Id12673df1797afe0af16812bdf67b8ca8a931626
Signed-off-by: Cristina Pauna <cristina.pauna@enea.com>

Merge "Auto Generated INFO.yaml file"

Auto Generated INFO.yaml file

Fix project lead

INFO.yaml meant to be machine readable.
This file was auto generated from an ldap lookup
and your projects INFO file.

Meeting times are not currently filled out.
please ammend this patchset to complete

Going forward comitter promotions should be done
against this file.

PTLs:
Must be merged for you to gain delete access on
testresults.opnfv.org

If project_lead: is missing (my script could not parse)
please amend with same from commiters section eg:

project_lead: &opnfv_dovetail_ptl
    name: 'Lead Name'
    email: 'lead email'
    id: 'lead lfid'
    company: 'lead company'

Cleaned up commiter list:
Jonas Bjurel – resigned
Guo, Ruijing – resigned
Stefan Berg – inactive last commit September 2016
Daniel Smith – inactive last commit June 2016
Szilard Cserey - inactive last commit Nov 2015
Michal Skalski: - inactive last commit August 2017
Nikolas Hermanns -inactive last commit  August 2016

Change-Id: I085ce063906f5596fdab0c9ece520ecb45c7e309
Signed-off-by: agardner <agardner@linuxfoundation.org>

[odl] Remove unused connecter/port from proxy

Change-Id: I7a880c0a011d6f5f4a7e2d8c8309871b1c290235
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>

[jump] Print CPU, RAM, disk info for debug

JIRA: FUEL-323

Change-Id: I74f92fbdf84762540401fe07a6eb424bdc7aa664
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>

[maas] Make boot-source-selection ops idempotent

- add `unless` clause;
- instead of adding a new boot sources, update the existing one,
  assuming standard MaaS installation (always updates id '1');

Change-Id: I2db92a3b14d0588095c69545cbb6977ca109613e
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>

Merge "Run galera state on slave nodes one by one"

Run galera state on slave nodes one by one

Apply galera state in consecutive order to avoid
a race condition with database initialization.

Change-Id: I877bad38777d8469c03cee3b7e96acc875a3a72a
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>

[maas] Set boot source selections

Select xenial image explicitly to avoid getting latest bionic one.

Change-Id: Iee872b8dd2ce309231829a0ad8e5a3cf1a75f796
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>

[submodule] Bump Pharos for zte-pod1 IDF update

Change-Id: I4942a1c3e12db304db11fbc231ca26ff87e3b6bb
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>

[docs] Updates for the first Fraser release

Change-Id: I80dfb55c01af89bded1140cd27563d258e05dda4
Signed-off-by: Cristina Pauna <cristina.pauna@enea.com>

Merge "[doc] Add guest OS support matrix"

[doc] Add guest OS support matrix

Update fuel documentation regarding the guest operating systems
that we can boot in both in x86 and aarch64.

Change-Id: Iea96fae880755d0a9cc8fa519362ff564f79e391
Signed-off-by: Charalampos Kominos <Charalampos.Kominos@enea.com>

[states] Catch more transient 'no response' resp

Change-Id: Ie8e60a648fa28e59daa6e00f357df52b5821e833
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>

Merge "Mend OVN scenario"

Update qemu/libvirt on compute nodes

Get qemu package with fix of TSC frequency mismatch between instance
and host which breaks up nova pause/unpause operation.

JIRA: FUEL-368
Change-Id: I2c954b2b0a8cddca30c861f565b63b642e0b0e9f
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>

Mend OVN scenario

* setup HWE kernel to get suitable conntrack module
* clean out outdated state with ovn ctl options
* point SB remote source to local mgmt network

Change-Id: I8986c227ce0a9a3b7ab3faf382760ec32e6e7c00
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>

[ha] Fix tenant NIC OS config for VLAN ranges

Previous commits added support for tenant VLAN range parametrization,
yet failed to accomodate the new VLAN range representation ('M:N') in
existing `linux.network.interfaces` reclass configuration.

This lead to the silent removal of <nic_private>.<vlan_private> from
interfaces file and for `br-mesh` to try to use <nic_private> instead
of <nic_private>.<vlan_private>, conflicting with `br-floating`.

Stick to using the first VLAN in the tenant range.

JIRA: FUEL-365

Change-Id: I88f7263159f1742dba3b6e36649c9344f2932c5f
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>

Disable cpu pinning & L3 HA features

* CPU pinning: tests have no support for dedicated CPU policy, so
  enabled pinning causes general performance impact
  [https://docs.openstack.org/nova/pike/admin/cpu-topologies.html]
* L3 HA routers:
  - doesn't catch l3-agent failure, relies only on
    HA network state to spot failure
  - failover process only retains the state of network
    connections for instances with a floating IP address
  - incompatible with DVR mode

JIRA: FUEL-360
Change-Id: Ie0182bf953b9989729f89d705d9fee902d229e51
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>

[reclass] Parameterize tenant VLAN ranges

Instead of hardcoding the VLAN ranges in reclass, read them from IDF.
This is required when VLAN segmentation is used instead of VXLAN, as
lab switches should be configured appropiately (e.g. Armband PODs use
VLANs 2000+ instead of 1000-1030).

Change-Id: I009ccea9b328c40f055a745791eafe3393b96390
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>

[baremetal] cmp linux.system: catch 'no response'

Catch & retry transient errors / timeouts while applying the
`linux.system` state on cmp nodes.

Change-Id: Id314b5a29673e0bcaa78611fc787491056830952
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>

[MaaS] destructive storage test fio on failure

Perform fio storage destructive test operation (usually takes just a
few minutes) to completely destroy any previous storage metadata that
might cause issues with cleanup in cloud-init/curtin during deploy.

Only resort to fio when a node fails to deploy, which allows us to
reuse the `maas.machines.mark_broken_state` state.

JIRA: FUEL-365

Change-Id: Ief327e6b4fefa83a8a3c131acfdf9f5fd605689d
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>

Build in opendaylight service mask

Oxygen has an issue with broken config/data cache caused
by service restart in the middle of initial boot.

Change-Id: Ia30c76b67566ab8a2fb9045d0e10ca788f1a06a6
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>