fuel.git
5 years agoSmooth down telemetry services 56/67256/1
Michael Polenchuk [Thu, 14 Mar 2019 15:08:39 +0000 (19:08 +0400)]
Smooth down telemetry services

* update gnocchi to 4.3
* remove outdated ceilometer api

Change-Id: I7adaf3ddc76d93531b6b0997b684672b80f2992f
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
5 years ago[lib] Create veths using systemd opnfv-fuel units 80/67180/1
Alexandru Avadanii [Tue, 5 Mar 2019 15:49:23 +0000 (16:49 +0100)]
[lib] Create veths using systemd opnfv-fuel units

Create 2 systemd services on the jumphost that will handle veth
pairs creation, respectively adding them to virsh/real bridges.
This allows us to set docker containers restart policy to 'always',
enabling persistent Salt Master/MaaS containers across jumphost
reboots.

NOTE: libvirt creates virtual networks async, hence the need for
retrying hooking veths to them.

JIRA: FUEL-406

Change-Id: I1ca033cb5eb854b577b57bb2387a58bd9605a5bb
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
5 years agoTurn off meltdown/spectre patches 69/67169/2
Michael Polenchuk [Mon, 4 Mar 2019 08:49:58 +0000 (12:49 +0400)]
Turn off meltdown/spectre patches

Change-Id: Id75ffe4db808a4ec250ba8b86c5d49f1206c3784
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
5 years agoTune up nova/neutron intervals 18/67118/3
Michael Polenchuk [Tue, 26 Feb 2019 14:09:14 +0000 (18:09 +0400)]
Tune up nova/neutron intervals

Also re-align resources for virtual scenarios.

Change-Id: Id0d55407fd5b1720a24e30c364219f8b08e89d06
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
5 years agoFix race condition with nova privsep utime 15/67115/1
Michael Polenchuk [Tue, 26 Feb 2019 10:52:06 +0000 (14:52 +0400)]
Fix race condition with nova privsep utime

Bug: https://bugs.launchpad.net/nova/+bug/1809123
Change-Id: I14622c21826aeeddac6ea7bf7f9d116cd3e68cfb
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
5 years agoMerge "[cfg01] Reduce mine_interval to 15 min"
Michael Polenchuk [Tue, 26 Feb 2019 07:03:04 +0000 (07:03 +0000)]
Merge "[cfg01] Reduce mine_interval to 15 min"

5 years ago[lib] Add fatal validation of old kernel on Ubuntu 86/67086/1
Alexandru Avadanii [Fri, 22 Feb 2019 15:31:24 +0000 (16:31 +0100)]
[lib] Add fatal validation of old kernel on Ubuntu

As reported in [1], kernel 4.4 seems to break nested virtualization,
add a fatal check against it.

[1] https://bugs.launchpad.net/ubuntu/+source/qemu/+bug/1797332

Change-Id: I0aef8a7340dd82bfeb2e58c9642623b9ec13dca5
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
5 years ago[cfg01] Reduce mine_interval to 15 min 66/67066/2
Alexandru Avadanii [Mon, 18 Feb 2019 22:08:30 +0000 (23:08 +0100)]
[cfg01] Reduce mine_interval to 15 min

Some PODs are fast enough to get past installing, syncing and using
MaaS to provision the OS on the baremetal nodes before the 1h mine
refresh.

Since mine.update operation is fast enough to go unnoticed and we
only collect IP addresses, grains and pem entries, schedule it every
15 minutes.

Due to reclass class inheritance, we can't easily override this via
pillar data, so handle it via entrypoint.sh.

Change-Id: I0d8ed2da838ad09c94e9327d0131d3e239de4f08
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
5 years agoInstall missing gnocchi dependencies 82/67082/1
Michael Polenchuk [Fri, 22 Feb 2019 08:45:52 +0000 (12:45 +0400)]
Install missing gnocchi dependencies

Change-Id: Ifc4fff90551344c69295990b220f0778967887a4
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
5 years agoMerge "[baremetal] Containerize MaaS"
Alexandru Avadanii [Tue, 19 Feb 2019 15:17:25 +0000 (15:17 +0000)]
Merge "[baremetal] Containerize MaaS"

5 years agoMerge "[cfg01] Schedule x509.get_pem_entries mine update"
Alexandru Avadanii [Fri, 15 Feb 2019 13:06:46 +0000 (13:06 +0000)]
Merge "[cfg01] Schedule x509.get_pem_entries mine update"

5 years ago[cfg01] Schedule x509.get_pem_entries mine update 33/67033/1
Alexandru Avadanii [Fri, 15 Feb 2019 00:23:20 +0000 (01:23 +0100)]
[cfg01] Schedule x509.get_pem_entries mine update

Previously, Salt Master CA mine was only sent once, during
salt.minion.ca state execution at cfg01 bringup / bootstrap.

This causes possible issues with:
- Salt Master container restart (mine data is lost);
- UNH Lab deployment (uknown rootcause, might be related to XFS and
  overlay2 being used with Docker on CentOS);

To bypass this issue, make x509.get_pem_entries module send mine data
at the default mine interval (60 minutes).

Change-Id: I5f6334ae18f5af6cbe0a164791603b67f0a3668f
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
5 years ago[baremetal] Containerize MaaS 99/66799/7
Alexandru Avadanii [Thu, 7 Feb 2019 18:51:04 +0000 (19:51 +0100)]
[baremetal] Containerize MaaS

- replace mas01 VM with a Docker container;
- drop `mcpcontrol` virsh-managed network, including special handling
  previously required for it across all scripts;
- drop infrastructure VMs handling from scripts, the only VMs we still
  handle are cluster VMs for virtual and/or hybrid deployments;
- drop SSH server from mas01;
- stop running linux state on mas01, as all prerequisites are properly
  handled durin Docker build or via entrypoint.sh - for completeness,
  we still keep pillar data in sync with the actual contents of mas01
  configuration, so running the state manually would still work;
- make port 5240 available on the jumpserver for MaaS dashboard access;
- docs: update diagrams and text to reflect the new changes;

Change-Id: I6d9424995e9a90c530fd7577edf401d552bab929
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
5 years agoRise up salt's gather job timeout 25/67025/1
Michael Polenchuk [Thu, 14 Feb 2019 10:33:38 +0000 (14:33 +0400)]
Rise up salt's gather job timeout

While the minions are working their jobs the CLI is waiting for the
first initial timeout period (timeout) to start. When that hits,
the CLI sends sends the first "find_job" query. This kicks off the
gather_job_timeout timer. Sometimes a minion doesn't respond to the request
within the gather_job_timeout time period (default is 10s), so rise up
this value to give a chance for a minion to report actual status.

Change-Id: Ic3756b82fdeb17718870ab30e9578263d25309f7
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
5 years agoMerge "[odl] Settle the broken pkg->config dependency"
Michael Polenchuk [Mon, 11 Feb 2019 11:32:22 +0000 (11:32 +0000)]
Merge "[odl] Settle the broken pkg->config dependency"

5 years agoMerge "[docker] Add MaaS container build support"
Alexandru Avadanii [Mon, 11 Feb 2019 10:42:24 +0000 (10:42 +0000)]
Merge "[docker] Add MaaS container build support"

5 years ago[odl] Settle the broken pkg->config dependency 99/66899/1
Michael Polenchuk [Fri, 8 Feb 2019 09:39:16 +0000 (13:39 +0400)]
[odl] Settle the broken pkg->config dependency

Change-Id: I3bbe3e4be520ccac198654bb4a7d493aa8450023
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
5 years ago[docker] Add MaaS container build support 95/66895/1
Alexandru Avadanii [Thu, 7 Feb 2019 18:50:33 +0000 (19:50 +0100)]
[docker] Add MaaS container build support

Change-Id: I7709c9ca9e701b656447154919eb084a710f49af
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
5 years ago[odl] Disable PaxOsgi logging by default 73/66873/3
Michael Polenchuk [Thu, 7 Feb 2019 08:36:56 +0000 (12:36 +0400)]
[odl] Disable PaxOsgi logging by default

The PaxOsgi logging has a performance impact
(i.e. makes pressure to the Java GC).

Change-Id: Ic0bc2c0d1cfac195a04d1cfa90fa7fa47fc37612
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
5 years agoMerge "[odl/noha] Make vif_plugging non-fatal"
Michael Polenchuk [Tue, 5 Feb 2019 07:34:08 +0000 (07:34 +0000)]
Merge "[odl/noha] Make vif_plugging non-fatal"

5 years agoMerge "[cfg01] Use ssh config to set default user & key"
Michael Polenchuk [Tue, 5 Feb 2019 07:32:32 +0000 (07:32 +0000)]
Merge "[cfg01] Use ssh config to set default user & key"

5 years agoMerge "[fdio] Fix VPP package pinning"
Alexandru Avadanii [Mon, 4 Feb 2019 21:33:27 +0000 (21:33 +0000)]
Merge "[fdio] Fix VPP package pinning"

5 years ago[fdio] Fix VPP package pinning 53/66853/1
Alexandru Avadanii [Mon, 4 Feb 2019 15:08:06 +0000 (16:08 +0100)]
[fdio] Fix VPP package pinning

Previously, Ubuntu ignored the VPP pinning with:

N: Ignoring file 'fdio.ubuntu' in directory '/etc/apt/preferences.d/'
as it has an invalid filename extension

Change-Id: I5ee60c1715bea3b4180b55125dc72962a70c2754
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
5 years ago[cfg01] Use ssh config to set default user & key 51/66851/1
Alexandru Avadanii [Mon, 4 Feb 2019 15:03:06 +0000 (16:03 +0100)]
[cfg01] Use ssh config to set default user & key

Change-Id: I7486569568207f7652f8bdfcf1060ce51a9dbb0e
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
5 years ago[submodule] Bump Pharos for arm-pod10 cmp change 49/66849/1
Alexandru Avadanii [Mon, 4 Feb 2019 14:08:20 +0000 (15:08 +0100)]
[submodule] Bump Pharos for arm-pod10 cmp change

Change-Id: Ia7f8845017333e54db110bca5b3715702948b76b
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
5 years ago[odl/noha] Make vif_plugging non-fatal 93/66793/2
Michael Polenchuk [Thu, 31 Jan 2019 12:34:28 +0000 (16:34 +0400)]
[odl/noha] Make vif_plugging non-fatal

In order to mitigate live migration procedure make VIF plugging
event non-fatal for nova-compute. Also align max value of memory
for instance of ODL controller.

Change-Id: I0d00cc97c652eef3bd3404fac4715e2e7f2f02c7
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
5 years agoMerge "[deploy] Allow only operating system install"
Cristina Pauna [Wed, 30 Jan 2019 10:29:53 +0000 (10:29 +0000)]
Merge "[deploy] Allow only operating system install"

5 years agoMerge "[fdio] Pin VPP packages to 18.07-release"
Cristina Pauna [Wed, 30 Jan 2019 08:43:03 +0000 (08:43 +0000)]
Merge "[fdio] Pin VPP packages to 18.07-release"

5 years ago[deploy] Allow only operating system install 77/66777/3
Alexandru Avadanii [Tue, 29 Jan 2019 17:05:44 +0000 (18:05 +0100)]
[deploy] Allow only operating system install

Extend one of the existing deployment arguments to allow the
installation of only the operating system and infrastructure networks,
skipping cloud setup.

Change-Id: Ibc5d0f324ed15b66f809839cfce49a0324b6fe4d
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
5 years agoMerge "[ovs] Start ovs services before networking"
Alexandru Avadanii [Tue, 29 Jan 2019 16:08:07 +0000 (16:08 +0000)]
Merge "[ovs] Start ovs services before networking"

5 years ago[fdio] Pin VPP packages to 18.07-release 71/66771/3
Alexandru Avadanii [Tue, 29 Jan 2019 15:15:50 +0000 (16:15 +0100)]
[fdio] Pin VPP packages to 18.07-release

VPP 18.10 has a weird bug triggered by certain packets, e.g. from
inside a guest VM on a compute node, these behave differently:
$ udhcpc -x hostname:1234567890123456789012  # works
$ udhcpc -x hostname:12345678901234567890123 # confuses VPP on gtw01

To avoid this bug, pin VPP to the previous release, which does not
exhibit the issue.

Change-Id: I8c1e085731909d4b9296e8b09608887a4b5bfdd6
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
5 years ago[fdio] Increase VIF plug-in timeout 77/66677/1
Alexandru Avadanii [Sun, 27 Jan 2019 19:15:02 +0000 (20:15 +0100)]
[fdio] Increase VIF plug-in timeout

Baremetal clusters might benefit from having a little more time
to plug in the VIFs.

Change-Id: I9406a0ef24de2177827b3acd27b7c60b293a4572
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
5 years ago[ovs] Start ovs services before networking 49/66649/5
Alexandru Avadanii [Fri, 25 Jan 2019 20:28:27 +0000 (21:28 +0100)]
[ovs] Start ovs services before networking

Fix broken systemd service unit dependecies:
- OVS should start before networking service;
- OVS ports & bridges should not be automatically ifup-ed by
  networking service to avoid races, so drop 'auto' for both
  (OVS ports are automatically handled when part of an OVS bridge);
- explicitly ifup OVS bridges as part of networking service, but
  after all Linux interfaces have been handled;
- use 'allow-ovs br-prv' to let OVS handle br-prv and avoid another
  race condition;

While at it, fix some other related issues:
- make OVS service start after DPDK service (if present);
- bump OVS-DPDK compute VMs RAM since since switching from MTU 1500
  to jumbo frames for virtual PODs a while ago failed to do so [1];
- avoid creating conflicting reclass linux.network.interfaces entries
  for OVS ports by using their name (drop 'ovs_port_' prefix):
  * for untagged networks they will override existing common defs;
  * for tagged networks, they will create separate entries;
- DPDK scenarios: make gtw01 br-prv members OVS ports to avoid race
  conditions after node reboot by letting OVS handle them;

[1] https://developers.redhat.com/blog/2018/03/16/\
    ovs-dpdk-hugepage-memory/

Change-Id: I0266ba67f3849b6f7e331a758146b331730bae55
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
5 years agoMerge "Enable back auto for ports network script"
Michael Polenchuk [Thu, 24 Jan 2019 11:55:07 +0000 (11:55 +0000)]
Merge "Enable back auto for ports network script"

5 years agoMerge "[fdio] Make VIF timeout non-fatal"
Alexandru Avadanii [Thu, 24 Jan 2019 10:53:27 +0000 (10:53 +0000)]
Merge "[fdio] Make VIF timeout non-fatal"

5 years agoEnable back auto for ports network script 91/66591/5
Michael Polenchuk [Wed, 23 Jan 2019 11:36:57 +0000 (15:36 +0400)]
Enable back auto for ports network script

The ovs port remains in down state after reboot if "auto" is off.
Also turn off no_wait option for odl-noha scenarios.

Change-Id: I0121b3190869528e5f2e9985f9e9299ac6c6724e
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
5 years ago[fdio] Make VIF timeout non-fatal 99/66599/1
Alexandru Avadanii [Thu, 24 Jan 2019 00:25:17 +0000 (01:25 +0100)]
[fdio] Make VIF timeout non-fatal

The first VMs spawned still exhibit the race condition described in
the ticket, so apply the same workaround proposed during the Fraser
release cycle in FDS.

JIRA: FDS-156

Change-Id: I3b2b1ed7b5711daf81b5f4a263e4dbee9f502259
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
5 years ago[docs] Update Gambia 7.2 release date 93/66593/1
Alexandru Avadanii [Wed, 23 Jan 2019 17:34:21 +0000 (18:34 +0100)]
[docs] Update Gambia 7.2 release date

Change-Id: I27d13cafcfa45f70413695dbb6fe29e5bb222a3e
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
5 years agoPass domain name properly for heat stack user 55/66555/2
Michael Polenchuk [Tue, 22 Jan 2019 08:53:07 +0000 (12:53 +0400)]
Pass domain name properly for heat stack user

Change-Id: I74c1c85310e2012e664764b6129fc4a52faaf106
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
5 years agoMerge "[uefi_cleanup] Use grain targeting"
Alexandru Avadanii [Mon, 21 Jan 2019 18:55:44 +0000 (18:55 +0000)]
Merge "[uefi_cleanup] Use grain targeting"

5 years agoMerge "[noha] baremetal: Fix undef armband_repo_version"
Michael Polenchuk [Mon, 21 Jan 2019 10:03:09 +0000 (10:03 +0000)]
Merge "[noha] baremetal: Fix undef armband_repo_version"

5 years ago[uefi_cleanup] Use grain targeting 05/66505/1
Alexandru Avadanii [Sat, 19 Jan 2019 20:10:49 +0000 (21:10 +0100)]
[uefi_cleanup] Use grain targeting

Alternating HA and no-HA scenario deployments on baremetal requires
non-hostname targeting for UEFI cleanup (e.g. ctl01/gtw01/kvm01).

Change-Id: I9f0e967b500856b65a69ea0ab6ea13e15b327d8b
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
5 years agoMerge "Sync heat domain name with keystone defined"
Alexandru Avadanii [Thu, 17 Jan 2019 17:38:09 +0000 (17:38 +0000)]
Merge "Sync heat domain name with keystone defined"

5 years ago[submodule] Bump Pharos for arm-pod10 cmp NIC sync 57/66457/1
Alexandru Avadanii [Thu, 17 Jan 2019 14:38:06 +0000 (15:38 +0100)]
[submodule] Bump Pharos for arm-pod10 cmp NIC sync

Change-Id: I177598d4d20539e50aab5f283e8d10022a4f1a14
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
5 years agoSync heat domain name with keystone defined 55/66455/1
Michael Polenchuk [Thu, 17 Jan 2019 13:38:23 +0000 (17:38 +0400)]
Sync heat domain name with keystone defined

Change-Id: Ibf88f179af2570a707ade78f772342b7da23b74f
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
5 years ago[noha] baremetal: Fix undef armband_repo_version 43/66443/1
Alexandru Avadanii [Wed, 16 Jan 2019 20:27:28 +0000 (21:27 +0100)]
[noha] baremetal: Fix undef armband_repo_version

Change-Id: I0e56261fc2fc2a0a3f164531c72d88f7c46f5ca1
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
5 years ago[submodule] Bump Pharos for arm-pod10 NIC reorder 25/66425/2
Alexandru Avadanii [Wed, 16 Jan 2019 12:38:27 +0000 (13:38 +0100)]
[submodule] Bump Pharos for arm-pod10 NIC reorder

Change-Id: I79d3167432d48500346d5c8294d447c54e0cb6be
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
5 years agoMerge "Align patches"
Michael Polenchuk [Wed, 16 Jan 2019 11:31:44 +0000 (11:31 +0000)]
Merge "Align patches"

5 years agoAlign patches 65/66365/2
Michael Polenchuk [Mon, 14 Jan 2019 13:18:50 +0000 (17:18 +0400)]
Align patches

* patch is merged into oslo-templates
* rocky repo key name is made as for others
* jinja package is updated to fix incorrect quoted value
  [https://github.com/saltstack/salt/issues/46594]

Change-Id: Ia6359cf89579b4d892ae40c4d087168edcd86ebb
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
5 years agoMake shutdown only on physical nodes 21/66421/2
Michael Polenchuk [Wed, 16 Jan 2019 10:03:48 +0000 (14:03 +0400)]
Make shutdown only on physical nodes

Change-Id: If167e7a6bdcdccd6b6df43bd5cac54250abec61a
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
5 years agoMerge "[baremetal] Shutdown nodes from previous deploy"
Alexandru Avadanii [Tue, 15 Jan 2019 13:52:05 +0000 (13:52 +0000)]
Merge "[baremetal] Shutdown nodes from previous deploy"

5 years ago[odl] Set conntrack as netvirt nat mode 37/66337/3
Michael Polenchuk [Fri, 11 Jan 2019 10:30:52 +0000 (14:30 +0400)]
[odl] Set conntrack as netvirt nat mode

The conntrack-based SNAT uses the Linux netfilter framework to
do the NAPT and track the connection. The first packet in a traffic is
passed to the netfilter to be translated with the external IP. The
following packets will use the netfilter for further inbound and
outbound translation.

Change-Id: I1090b4fe041f8d9533aa4ce1964284a4a5c073ce
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
5 years agoMerge "[centos] Update altarch kernel URL"
Michael Polenchuk [Mon, 14 Jan 2019 08:57:33 +0000 (08:57 +0000)]
Merge "[centos] Update altarch kernel URL"

5 years agoMerge "[patch] Drop reclass.system patch for repo arch"
Michael Polenchuk [Mon, 14 Jan 2019 08:55:39 +0000 (08:55 +0000)]
Merge "[patch] Drop reclass.system patch for repo arch"

5 years agoMerge "[noha] Fix gtw private NIC name in j2 templates"
Michael Polenchuk [Mon, 14 Jan 2019 08:46:11 +0000 (08:46 +0000)]
Merge "[noha] Fix gtw private NIC name in j2 templates"

5 years ago[baremetal] Shutdown nodes from previous deploy 51/66351/1
Alexandru Avadanii [Sun, 13 Jan 2019 17:49:07 +0000 (18:49 +0100)]
[baremetal] Shutdown nodes from previous deploy

When noha scenarios are scheduled on the same CI POD currently
running a previously deployed HA scenario, one baremetal node
might remain unused (kvm03), connect to the new Salt master and
interfere with the deployment.

To prevent that, shutdown all baremetal nodes at the begining of the
deployment.

Change-Id: Ia9bad8b5d8348433cefac9aa76eca0de664f187d
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
5 years ago[centos] Update altarch kernel URL 27/66127/2
Alexandru Avadanii [Mon, 31 Dec 2018 19:43:51 +0000 (20:43 +0100)]
[centos] Update altarch kernel URL

CentOS recently moved its kernel source RPM from the altarch subdir
to the same directory x86_64 kernel sources used to reside, so update
our script accordinly.

Change-Id: I88010eabdfc15d6a79350dface29258cc37c4b95
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
5 years ago[patch] Drop reclass.system patch for repo arch 21/66321/2
Alexandru Avadanii [Thu, 10 Jan 2019 14:47:00 +0000 (15:47 +0100)]
[patch] Drop reclass.system patch for repo arch

MCP repos no longer publish arm64 metadata, so drop our patch that
selected arm64 metadata on arm64 systems.
Instead, let it default to 'deb [arch=amd64]', which will allow
arm64 systems to fetch amd64 metadata and inherintely fetch all
arch-independent packages from the same repos.

While at it, switch to 'rocky-armband' repos on arm64 systems.

Change-Id: I07fda895f5162bfa576c62336cbb4d74e985f37a
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
5 years ago[noha] Fix gtw private NIC name in j2 templates 43/66343/1
Alexandru Avadanii [Fri, 11 Jan 2019 18:38:41 +0000 (19:38 +0100)]
[noha] Fix gtw private NIC name in j2 templates

Change-Id: Ic266864913dcac021b3e12f426e1c8a60c23fe87
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
5 years ago[patch] Avoid ifup run if noifupdown is turned on 11/66111/8
Michael Polenchuk [Tue, 25 Dec 2018 10:28:47 +0000 (14:28 +0400)]
[patch] Avoid ifup run if noifupdown is turned on

Handle noifupdown option for all cmd.run states
with explicit ifup call as well.

Change-Id: Ie855a0810bcfe4a856cf9d29bd0755643d71ff4d
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
5 years agoMerge "[docs] Update documentation for Gambia 7.2"
Alexandru Avadanii [Thu, 10 Jan 2019 14:25:51 +0000 (14:25 +0000)]
Merge "[docs] Update documentation for Gambia 7.2"

5 years agoMerge "[state] Fold aarch64 conditions"
Alexandru Avadanii [Thu, 10 Jan 2019 14:07:23 +0000 (14:07 +0000)]
Merge "[state] Fold aarch64 conditions"

5 years ago[docs] Update documentation for Gambia 7.2 17/66317/1
Cristina Pauna [Thu, 10 Jan 2019 12:32:45 +0000 (14:32 +0200)]
[docs] Update documentation for Gambia 7.2

Change-Id: I180f668b297ad97dd95bd9201005410fe7a62b4c
Signed-off-by: Cristina Pauna <cristina.pauna@enea.com>
5 years ago[state] Fold aarch64 conditions 09/66309/1
Alexandru Avadanii [Wed, 9 Jan 2019 14:47:47 +0000 (15:47 +0100)]
[state] Fold aarch64 conditions

The armband formula already has checks in place to run only on
nodes with the expected arch, so remove the duplicate condition
in state files.

Change-Id: I05b26368a2d97422830a692e09242bc50e4eb1db
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
5 years agoBring in FDIO (VPP+DPDK) scenario 71/66271/5
Alexandru Avadanii [Thu, 8 Nov 2018 18:06:46 +0000 (19:06 +0100)]
Bring in FDIO (VPP+DPDK) scenario

- cmp, gtw: bump RAM allocation to accomodate hugepages/VPP;
  for now we overcommit, gtw01 resources can probably be lowered;
- submodule: add salt-formula-neutron so we can locally patch it;
- repo:
  * FD.IO repos for VPP packages;
  * networking-vpp PPA for python-networking-vpp Neutron driver;
- use vpp-router for L3, disable neutron-l3-agent;
- baremetal_init: apply repo config before network (otherwise UCA
  repo is missing when trying to install DPDK on baremetal nodes);
- arm64: iommu.passthrough=1 is required on ThunderX for VPP on
  newer kernels;

Design quirks:
- vpp service runs as 'neutron' user, which does not exist at the
  time VPP is installed and initially started, hence the need to
  restart it before starting the vpp-agent service;
- gtw01 node has DPDK, yet to configure it via IDF we use the
  compute-specific OVS-targeted parameters like
  `compute_ovs_dpdk_socket_mem`, which is a bit misleading;
- vpp-agent requires ml2_conf.ini on ALL compute AND network nodes
  to parse per-node physnet-to-real interface names;
- vpp process is bound to core '1' (not parameterized via IDF);

Change-Id: I659f7dbebcab7b154e7b1fb829cd7159b4372ec8
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
5 years agoMerge "[MaaS] Implement aarch64 tags for kernel_opts"
Alexandru Avadanii [Thu, 3 Jan 2019 15:17:27 +0000 (15:17 +0000)]
Merge "[MaaS] Implement aarch64 tags for kernel_opts"

5 years ago[MaaS] Implement aarch64 tags for kernel_opts 15/66115/4
Alexandru Avadanii [Sat, 29 Dec 2018 22:54:32 +0000 (23:54 +0100)]
[MaaS] Implement aarch64 tags for kernel_opts

On AArch64, 1G hugepages need to be enabled via kernel cmdline
before mounting hugetlbfs [1].
Leverage MaaS tags to apply custom kernel args to AArch64 nodes.

[1] https://wiki.debian.org/Hugepages

Change-Id: Ie68ddf805836ee62f725019b0b873082b1d40948
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
5 years agoRectify tacker server version 09/66109/1
Michael Polenchuk [Sat, 29 Dec 2018 11:02:44 +0000 (15:02 +0400)]
Rectify tacker server version

Change-Id: I270660204653d06cc8d1b5dc773d11a0a05ac27b
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
5 years agoRemove rabbitmq hostname option 53/66053/3
Michael Polenchuk [Tue, 25 Dec 2018 10:28:47 +0000 (14:28 +0400)]
Remove rabbitmq hostname option

The BindToDevice=lo option has been removed from epmd.socket:
https://bugs.launchpad.net/ubuntu/+source/rabbitmq-server/+bug/1808766

Change-Id: Ib92eaf5efababd37963745bdb8e3ad9354045449
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
5 years agoMerge "[cfg01] salt-master: Sync cfg from reclass"
Michael Polenchuk [Tue, 25 Dec 2018 11:43:24 +0000 (11:43 +0000)]
Merge "[cfg01] salt-master: Sync cfg from reclass"

5 years agoMerge "[armband] Parameterize aarch64 specific repo comp"
Michael Polenchuk [Tue, 25 Dec 2018 11:42:53 +0000 (11:42 +0000)]
Merge "[armband] Parameterize aarch64 specific repo comp"

5 years ago[ha] Fix pillar data targeting msg nodes 49/66049/1
Alexandru Avadanii [Mon, 24 Dec 2018 12:29:34 +0000 (13:29 +0100)]
[ha] Fix pillar data targeting msg nodes

Change-Id: If4f4eb08c54ba8265ac178f215384d8a749d1b8b
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
5 years ago[armband] Parameterize aarch64 specific repo comp 39/66039/1
Alexandru Avadanii [Sun, 23 Dec 2018 15:45:43 +0000 (16:45 +0100)]
[armband] Parameterize aarch64 specific repo comp

Factor out armband repository component name based on
openstack_version parameter. However, since 'rocky-armband' repo
is not available yet, reuse the 'queens-armband' on.

Change-Id: Ibd17808059542ace14724776f8686fc2afd9de6a
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
5 years ago[patch] Fix whitespace at EOL 35/66035/1
Alexandru Avadanii [Sat, 22 Dec 2018 19:55:44 +0000 (21:55 +0200)]
[patch] Fix whitespace at EOL

`make patches-export` trims trailing whitespace from patch file,
bring it back to prevent issues with `patches-import`.

Change-Id: Ie9fe31f4480164ce19d3ccd47c2050e28382410a
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
5 years agoPull out rocky patches 03/66003/8
Michael Polenchuk [Thu, 20 Dec 2018 11:32:48 +0000 (15:32 +0400)]
Pull out rocky patches

This commit should be reverted once original formulas
get required support of rocky version.

Change-Id: Ia3458381bced0cae8dbfacc9781c90933ad5c822
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
5 years ago[cfg01] salt-master: Sync cfg from reclass 85/65985/1
Alexandru Avadanii [Sat, 15 Dec 2018 00:26:44 +0000 (01:26 +0100)]
[cfg01] salt-master: Sync cfg from reclass

Sync predefined Salt Master configuration with reclass class
system.salt.master.single (but limit worker_threads to 20).

Change-Id: I760cdcb9ebbdab517011eccab0616abb36014cc1
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
5 years agoUpdate OpenStack version to Rocky 61/65361/3
Michael Polenchuk [Fri, 23 Nov 2018 13:17:59 +0000 (17:17 +0400)]
Update OpenStack version to Rocky

Change-Id: I88f28370180278c4b32599b83eebbb0ee005c936
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
5 years ago[mas01] Fix iptables pillar compatibility format 25/65925/1
Alexandru Avadanii [Mon, 17 Dec 2018 18:17:59 +0000 (19:17 +0100)]
[mas01] Fix iptables pillar compatibility format

Sync our reclass pillar data for mas01's iptables with latest formula
changes [1].

[1] https://github.com/salt-formulas/salt-formula-iptables/commit/e353ce3c

Change-Id: I66b2a75066ed512ab5ab4cc213d13d15c5c8cc7f
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
5 years ago[maas] Adopt maas, maasng proposed functions 07/62807/13
Alexandru Avadanii [Sat, 22 Sep 2018 19:54:01 +0000 (21:54 +0200)]
[maas] Adopt maas, maasng proposed functions

JIRA: FUEL-364

Change-Id: Ia470fc8103713e7a06cd9647675b0edfb4342bf8
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
5 years ago[docker] deploy.sh: Use tag for stable branch 33/65833/1
Alexandru Avadanii [Fri, 14 Dec 2018 14:25:40 +0000 (15:25 +0100)]
[docker] deploy.sh: Use tag for stable branch

Outside OPNFV Jenkins (i.e. when manually cloning the OPNFV Fuel
repo and starting a deploy), the Docker tag used to default to
'latest' unless the user specifically set it to 'gambia'.
Rely on 'defaultbranch' setting in .gitreview to determine the
appropiate Docker tag.

Change-Id: I7e6b0706597d84d7cd5dc077499da78031aa61af
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
5 years agoMerge "[docs] Updates for Gambia 7.1.0 release"
Alexandru Avadanii [Fri, 14 Dec 2018 13:29:05 +0000 (13:29 +0000)]
Merge "[docs] Updates for Gambia 7.1.0 release"

5 years agoMerge "Documentation for ONAP scenarios"
Alexandru Avadanii [Fri, 14 Dec 2018 13:11:05 +0000 (13:11 +0000)]
Merge "Documentation for ONAP scenarios"

5 years agoMerge "[VCP] Reboot all VCP VMs after network config"
Alexandru Avadanii [Thu, 13 Dec 2018 15:08:27 +0000 (15:08 +0000)]
Merge "[VCP] Reboot all VCP VMs after network config"

5 years ago[docs] Updates for Gambia 7.1.0 release 91/65791/4
Alexandru Avadanii [Thu, 13 Dec 2018 14:05:35 +0000 (15:05 +0100)]
[docs] Updates for Gambia 7.1.0 release

While at it, rename FDIO (VPP) scenarios to align with OPNFV FDS
and OPNFV Apex projects.

Change-Id: I9aab5dc4a0dc41a2cc996687a8a2726d03288678
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
5 years agoDocumentation for ONAP scenarios 85/65785/2
Richard Elias [Thu, 13 Dec 2018 10:56:37 +0000 (11:56 +0100)]
Documentation for ONAP scenarios

Patch adds a short description for the two ONAP scenarios added by
JIRA: AUTO-71(os-nosdn-onap-*ha).

JIRA: AUTO-90

Change-Id: I0a25b1877090a16d365e06dc41705239ada9d0b0
Signed-off-by: Richard Elias <richard.elias@tieto.com>
5 years agoMerge "Make MTU cluster-level configurable via IDF"
Alexandru Avadanii [Wed, 12 Dec 2018 19:44:34 +0000 (19:44 +0000)]
Merge "Make MTU cluster-level configurable via IDF"

5 years ago[VCP] Reboot all VCP VMs after network config 59/65759/1
Alexandru Avadanii [Wed, 12 Dec 2018 19:37:27 +0000 (20:37 +0100)]
[VCP] Reboot all VCP VMs after network config

Currently, PXE/admin on VCP nodes still uses MaaS DHCP before
the new network configuration is enforced. A live network
refresh would break minion registration with the Salt master,
so perform a node reboot instead.

Change-Id: I1c25f63f6c98a9fff98108d3fad9550dd4468355
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
5 years agoMake MTU cluster-level configurable via IDF 37/65737/6
Alexandru Avadanii [Tue, 11 Dec 2018 14:56:21 +0000 (15:56 +0100)]
Make MTU cluster-level configurable via IDF

JIRA: FUEL-336

Change-Id: I1c8d22b8322f700eb727d9077035ba4c9f9f9753
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
5 years ago[docs] Update scenario matrix: os-odl-bgpvpn-noha 65/65365/2
Stamatis Katsaounis [Fri, 23 Nov 2018 14:11:35 +0000 (16:11 +0200)]
[docs] Update scenario matrix: os-odl-bgpvpn-noha

Scenario description resides in the SDNVPN repository.

JIRA: FUEL-393

Change-Id: I2c4f3c32821ed47ea104706bd1bff6e2e55eb50f
Signed-off-by: Stamatis Katsaounis <mokats@intracom-telecom.com>
5 years agoMerge "[bgpvpn] Use Linux bridge for odl01 public network"
Alexandru Avadanii [Mon, 10 Dec 2018 13:55:12 +0000 (13:55 +0000)]
Merge "[bgpvpn] Use Linux bridge for odl01 public network"

5 years ago[odl/dpdk] Disable up/down for public interface 91/65691/1
Michael Polenchuk [Mon, 10 Dec 2018 09:00:07 +0000 (13:00 +0400)]
[odl/dpdk] Disable up/down for public interface

There is a race with interfaces up/down action during
configuration, so activate them after node reboot.

Change-Id: Id40ce746cc6635fcedd0f9c809cf4a9fe4d1f034
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
5 years agoMerge "[ha] barbican, nova targeting: use server:role"
Michael Polenchuk [Mon, 10 Dec 2018 07:03:45 +0000 (07:03 +0000)]
Merge "[ha] barbican, nova targeting: use server:role"

5 years ago[bgpvpn] Use Linux bridge for odl01 public network 87/65687/1
Alexandru Avadanii [Sun, 9 Dec 2018 14:34:52 +0000 (15:34 +0100)]
[bgpvpn] Use Linux bridge for odl01 public network

Previously, we used a single interface definition for public network
on odl01 node, which does not work well for baremetal setups that
use a tagged VLAN public network, like ericsson-pod1.

Change-Id: I10ff7c105406691011e94e06b2f099dc2cdf8a06
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
5 years ago[cfg01] reclass: Apply broken default patch first 73/65673/1
Alexandru Avadanii [Sat, 8 Dec 2018 16:02:37 +0000 (17:02 +0100)]
[cfg01] reclass: Apply broken default patch first

reclass settings.py should be patched before the salt-master service
is started (since we can't restart it inside the container without
killing the tini init).

Fixes: 2de5348a

Change-Id: Id62d8f9f12fd72ef60322dd9907f26907231c4a7
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
5 years ago[ha] barbican, nova targeting: use server:role 57/65657/2
Alexandru Avadanii [Fri, 7 Dec 2018 14:57:13 +0000 (15:57 +0100)]
[ha] barbican, nova targeting: use server:role

Instead of matching the hostname to determine the primary barbican
node, rely on pillar data.

While at it, apply the same logic to nova:controller states.

Change-Id: Ia4ac391927a8b4b223febff106b400e84d264f7a
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
5 years ago[cfg01] Patch reclass settings.py broken default 51/65651/1
Alexandru Avadanii [Fri, 7 Dec 2018 14:42:23 +0000 (15:42 +0100)]
[cfg01] Patch reclass settings.py broken default

Until a new reclass release including the fix for [1] is published,
patch settings.py explicitly.

[1] https://github.com/salt-formulas/reclass/issues/77

Change-Id: If93848a6b6f084c76e64ddcf7bfdfa52b7d59050
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
5 years ago[ha] fix barbican server install error 35/65635/2
Guillermo Herrero [Fri, 7 Dec 2018 08:09:38 +0000 (09:09 +0100)]
[ha] fix barbican server install error

Ensure installation happens first on node 1 first,
run as batch on the rest after.

Change-Id: I1ac3f516c35c4941b343e88d47c78b19d13d7b90
Signed-off-by: Guillermo Herrero <guillermo.herrero@enea.com>
5 years agoMerge "Initial implementation of ONAP scenarios"
Michael Polenchuk [Thu, 6 Dec 2018 12:37:11 +0000 (12:37 +0000)]
Merge "Initial implementation of ONAP scenarios"

5 years agosubmodule: Bump pharos to latest 65/65565/1
Paul Vaduva [Tue, 4 Dec 2018 13:12:59 +0000 (15:12 +0200)]
submodule: Bump pharos to latest

Change-Id: I1bf4452e0f6e9aa5d2b9a002a1ec45c70fb8c2ab
Signed-off-by: Paul Vaduva <Paul.Vaduva@enea.com>
5 years agoInitial implementation of ONAP scenarios 69/64369/51
Martin Klozik [Thu, 1 Nov 2018 08:49:45 +0000 (09:49 +0100)]
Initial implementation of ONAP scenarios

Patch introduces a new specific scenarios os-nosdn-onap-*ha
for automatic ONAP deployment on top of the OPNFV installation.
Deployment and configuration of ONAP is managed by OPNFV Auto project.

New scenarios are based on generic os-nosdn-nofeature-*ha scenarios.

Auto project is responsible for os-nosdn-onap-*ha development
and maintenance.

JIRA: AUTO-71

Change-Id: I8b177668d856f30b62d1d135b80a95c32ebb9937
Signed-off-by: Martin Klozik <martin.klozik@tieto.com>