Age | Commit message (Collapse) | Author | Files | Lines |
|
Change-Id: I55a3c10f275079b11b7456b28a2c846cb33c204a
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
|
|
- adopt NOTE tags;
- switch some NOTE tags to WARNING;
- use ``markup`` for all paths;
Change-Id: If1bda281fd5cbef0b94110b59e02f41c443b0fb9
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
|
|
Restart docker service to refresh FORWARD chain
and insert docker related rules on top.
Change-Id: I971840f5979636c4ea8ae4d66a82982c24aa5f66
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
|
|
JIRA: FUEL-385
After successfully merging patch
https://gerrit.opnfv.org/gerrit/#/c/61417/ a user can deploy
os-nosdn-nofeature-noha scenario with numa topologies on compute nodes.
In order to be able to run the EPA testcases, there are still some
requirements (see [1]):
- Testcases expect compute nodes to have hugepages enabled
- Testcases expect compute nodes to have cgroup-tools package
installed
- Testcases expect to have NUMATopologyFilter added to
enabled_filters in /etc/nova/nova.conf of the controller node
This patch tries to meet the above requirements. First, it installs
the cgroup-tools package on compute nodes. Secondly, it overrides the
default enabled_filters of openstack nova salt package by appending
NUMATopologyFilter in the end (see [2], [3]).
Reference links:
[1] https://docs.google.com/document/d/1sT63M6fnurn4rgYTiUd8ILuXORtx0oHA8Qe-nPeCr2Y/edit
[2] https://github.com/salt-formulas/salt-formula-nova/blob/master/README.rst#custom-scheduler-filters
[3] https://github.com/salt-formulas/salt-formula-nova/blob/master/metadata/service/control/single.yml#L20
Change-Id: I24eb86c53574c80ceb33ecd1bfcb9ef2727d4263
Signed-off-by: Stamatis Katsaounis <mokats@intracom-telecom.com>
|
|
Each compute node has 16 GB of RAM but those GBs are splited wrongly
to each NUMA node (4 GB instead of 8 GB).
JIRA: FUEL-387
Change-Id: Ibe784727c8a262b41add04ad3a41c70f25ded684
Signed-off-by: Dimitrios Markou <mardim@intracom-telecom.com>
|
|
* to be compatible with HWE kernel
* make neutron balancing to RR back
* turn off glance v1 api support (doesn't relevant for now)
Change-Id: I916aae10f523be339c20de32218ce03c245afe72
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
|
|
JIRA: FUEL-383
Change-Id: I9203aa8d20def5b78d261f8c6847ddc576f0feb7
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
|
|
|
|
|
|
Added/updated the scenario files for the scenarios we support in Gambia
Change-Id: Ib463e3122999e0cf554599fb8637c20cf77b825d
Signed-off-by: Cristina Pauna <cristina.pauna@enea.com>
|
|
Since we switched to dockerized Salt master and mounting the full
contents of the current git repo inside cfg01's /root/fuel dir,
exposing temporary deploy artifacts should be avoided, i.e. the
default temporary storage location (configurable via '-S' deploy arg)
should be moved outside the git repo.
JIRA: FUEL-383
Change-Id: I32f7197b018fd853867de42f5618650ac9022dc9
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
|
|
While at it, retire obsolete MAAS_IP global variable and let mas01
VM get a DHCP address from virsh-managed mcpcontrol network.
Change-Id: Ifd85dbcab10894a5d0d675d37f0c35f09776d9b4
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
|
|
Change-Id: Ie8ec7c94b2831dce88bde39c7fe219faaad21c5e
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
|
|
|
|
|
|
Install in a local directory a newer version of virt-manager
to workaround obsolete Ubuntu versions lacking --cpu cellN.* support.
This change only affects CPU cfg of virtual compute nodes in
nosdn-nofeature-noha scenarios with:
- set default cpu_topology to dual socket (2 cores, 2 sockets,
2 threads);
- bump default RAM to 16GB;
- define 2 NUMA cells, each with half the resources;
To keep the old behavior available (single socket), a new deploy
argument has been added (`-m`). The RAM change is not configurable
via deploy args.
NOTE: The CPU topology for virtual nodes should later be read from
PDF instead of hardcoding it on a per-scenario basis in the installer.
NOTE: Default 'ram' unit is MiB, while cellN.memory default unit is
'KiB'.
JIRA: FUEL-385
Change-Id: I7ca268b0a2052524cb7187a5cf9b6fa8a382c9f9
Signed-off-by: Dimitrios Markou <mardim@intracom-telecom.com>
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
|
|
Allow skipping docker pull for verify jobs by setting the new env
var to 'verify'.
JIRA: FUEL-383
Change-Id: If8e2f66b5ccdac5c3911eeabfc2ba9c0eba61093
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
|
|
`system.neutron.control.cluster`, unlike
`system.neutron.control.openvswitch.cluster` does not provide
neutron:server:message_queue:members pillar data, letting it
default to the wrong values (single rabbit instance on the same node).
This led to neutron.conf on ctl nodes using:
`transport_url=rabbit://openstack:***@172.16.10.35:5672//openstack`
instead of
`transport_url=rabbit://openstack:***@172.16.10.28:5672,...//openstack`
Change-Id: Iad4b709d555b2bafafeb75fdecb831f7d4f5a504
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
|
|
|
|
Change-Id: I7d16bcd42a059817d7a4e6b06490e03001354f4f
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
|
|
|
|
Drop duplicate maas:machines definitions which could cause conflicts
in rare corner cases.
Slightly refactor j2 template expansion to make `conf.virtual.nodes`
available during first stage.
Change-Id: I04d56e346b12c6eb97da5c0c0ab1e3446e5fc1b8
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
|
|
Change-Id: I5aad512b5761deaa30522b6939bf245ecb1ec681
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
|
|
Change-Id: I023b7f47245691fe7377e53ca0bda15098c2978a
Signed-off-by: Cristina Pauna <cristina.pauna@enea.com>
|
|
|
|
openstack/control.yml
-redefine database host ip inherited from
system.neutron.control.cluster
Change-Id: Ic8e61c61c7ebb17f31e0c53f8d9e3a013f8a3e9e
Signed-off-by: Guillermo Herrero <guillermo.herrero@enea.com>
|
|
Change-Id: I448aa7f076e2c2d641a5326947852b3333728460
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
|
|
|
|
|
|
Enable jumbo frames on tenant/private interface as well so that
an instance can get the DHCP response from gateway node.
Change-Id: I9407fa67e9e0cdbe50335e4430748b0e45ba841a
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
|
|
|
|
In order to avoid TCP connection checksum issue (i.e. TX offloading
on ovs bridges) add linux bridge connected with ovs public bridge.
Change-Id: I4d266dd92756d5326dfa3d74fe2f376b26415812
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
|
|
JIRA: FUEL-383
Change-Id: Ie635965cff3538d6e786f2b76fa4d175d1be1773
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
|
|
Make the bulk of scenario files static again by shifting out all
common virtual nodes (mas01) and states (virtual_init, maas etc.)
to default.yaml(.j2).
This allows us to parse scenario-specific data during first j2
expansion, preparing for the new Pharos installer adapter that
relies on `conf.virtual.nodes.control` length to construct the
proper list of MaaS node definitions (kvm{01,02,03} vs {ctl01,gtw01}).
Change-Id: I666ab5bd6bb2a42f98646af51950f6b9fffa0e8b
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
|
|
* ship prebuilt salt master conf for better readability:
- enable x509.sign_remote_certificate (for prx VCP nodes);
* refactor Salt master CA handling:
- preinstall `salt_minion_dependency_packages` and
`salt_minion_reclass_dependencies` inside docker image;
- persistent /etc/pki;
- run salt.minion on cfg01 to generate master keys;
* bump container formulas to 1 Sep 2018 versions or newer:
- inject date into Docker makefile, forcing a fresh fetch of all
salt formulas from upstream git repos;
* workaround broken salt-formula-designate's meta/sphinx.yml:
- the DEB package version of salt-formula-designate uses `cmd.shell`
to query dpkg on the minion, while the git repo version still
uses `cmd.run`, running into parsing issues;
- temporarily disable sphinx metadata generation for designate until
upstream git repo syncs with the DEB version;
* upstream: salt-formula-salt AArch64 salt.control.virt support:
- retire salt-formula-salt git submodule and related patches;
* skip installing reclass distro package (already installed via pip
inside the container);
* limit initial pillar_refresh call to nodes on jumphost;
* remove unused salt-formula-nova git submodule;
JIRA: FUEL-383
Change-Id: I883b825e556f887a5e31f8a43676dcd8ece6dfde
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
|
|
JIRA: FUEL-404
Change-Id: I12acc90cdb17ecc090a993e1be41d13ad81dbb9d
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
|
|
JIRA: FUEL-383
Change-Id: Ib01e78951403fbec2e0f588ae5a577bcbad1bd75
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
|
|
While at it, use explicit `docker-compose pull` call to maintain
backwards compatibility with older docker-compose (1.8.0 on aarch64).
JIRA: FUEL-383
Change-Id: I0f6ac4fb59c0a286ac99a1a0dc4484d00f3837fe
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
|
|
* shift MTU from public bridge to physical interface
* add neutron related settings
Change-Id: Ia57d1ca7976968d6e7ee23f58a0abae1a1a256c0
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
|
|
Salt relies on a limiting libvirt_domain j2 template to generate the
XML it passes to libvirt for salt.control managed virtual machines.
For AArch64, we need to set up 3 XML nodes in a non-default way:
1. UEFI firmware (AAVMF) should be enabled by passing a pflash loader;
2. CPU mode should be 'host-passthrough';
3. QEMU machine type should be 'virt';
To allow configuring the above using pillar data:
- virtng module: implement functionality similar to upstream changes:
* 219b84a512 virt module: Allow NVRAM unlinking on DOM undefine
in develop, not in 2018.2;
* 9cace9adb9 Add support to virt for libvirt loader
in develop, not in 2018.2;
- virtng module: extend it with:
* pass virt machine type to vm;
* pass cpu_mode to vm;
JIRA: ARMBAND-404
Change-Id: Ib2123e7170991b3dfbdb42bd1a2baa5a4360b200
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
|
|
curtin generates armband_3.list, which we aligned in the reclass
model before, but not in the defaults yml used by cfg01/VCP VMs.
Change-Id: Ibe8d5b79465b508e41c20ec08a98a96a1510da8d
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
|
|
Since we reboot all nodes, applying the network configuration via
Salt before reboot is pointless and creates a race condition with
OVS.
While at it, add `--ignore-errors` to ifup call for OVS bridge to
prevent a race condition during linux.network state apply.
Change-Id: I22fe0afaffecd7b850a6b77d7b810ed296bfc9ca
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
|
|
AArch64 specific formula, mostly tweaking nova conf / installing
virtualization layer prerequisites:
- install qemu-efi;
- install vgabios;
- fix missing link for vgabios binary blob;
- nova conf: cpu_model=cortex-a57 (only for virtual deploys);
- nova conf: virt_type=qemu (only for virtual deploys);
- nova compute conf: virt_type=qemu (only for virtual deploys);
- nova conf: pointer_model=ps2mouse since AArch64 has no USB tablet;
[1] https://github.com/openstack/nova/commit/f0f0953
Change-Id: I40515bdbd941850b103a86d51b347cc8610f5741
Signed-off-by: Guillermo Herrero <Guillermo.Herrero@enea.com>
Signed-off-by: Charalampos Kominos <Charalampos.Kominos@enea.com>
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
|
|
* refactor & extend `-f` deploy argument;
* retire INFRA_CREATION_ONLY env var, duplicated by
NO_DEPLOY_ENVIRONMENT;
* `-F` and `-e` deploy arguments are now equivalent;
JIRA: FUEL-383
Change-Id: Ifc1527fa1e7d7486d1b7600772e2c5de34b1e52c
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
|
|
Salt bootstrap scripts are no longer used directly, so it is now safe
to retire the whole git submodule and its related patches.
JIRA: FUEL-383
Change-Id: I1fbdfe4fbd4930bfb3c999a3a68033d12565682b
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
|
|
* Refactor OPNFV salt-formulas mechanism to resemble upstream git
structure:
- git submodules: add new submodule for each formula we patch;
- create salt-formula-x directories for OPNFV formulas;
- move mcp/metadata/service contents to their each formula subdir;
- use `make patches-import` for patches previously handled by
patch.sh;
- retire patch.sh
* states: add virtual_init:
- mostly based on old salt.sh, which is now obsolete;
- exclude salt-master service restart (it would kill the container);
* scenarios: cleanup (rm cfg01 virtual node def), adopt virtual_init;
* reclass: align our model with prebuilt container's Salt config:
- drop linux:network pillar data (handled by Docker);
- stop applying linux.system state on cfg01;
- align salt user homedir;
- drop salt-formula packages (preprovisioned);
* minor plumbing in deploy.sh and lib.sh;
JIRA: FUEL-383
Change-Id: I28708a9b399d3f19012212c71966ebda9d6fc0ac
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
|
|
Add configuration items for bringing up the Salt master Docker
container using docker-compose:
- mount <storage dir>/hosts as /etc/hosts
- mount <storage dir>/salt as /etc/salt:
* semi-persistent generated Salt information;
- mount <storage dir>/nodes as /srv/salt/reclass/nodes:
* semi-persistent generated reclass node data storage;
- mount Fuel@OPNFV git repo in /root/fuel;
- mount all other previously rsynced artifacts;
- hook container to 3 networks/bridges: mcpcontrol, PXE/admin and
management:
* container IP addresses are managed by Docker, so no DHCP is used;
- override upstream entrypoint.sh by mounting our own version on top;
* run-time patch Salt to fix certain init system interactions due
to 'Tini' looking like Upstart but behaving a bit differently;
* work around directory traversal issues in Salt/reclass by using
extensive `cp` (to be later mitigated at container build time);
* overwrite reclass.system with our own version (patched), later to
fetched pre-patched during container build;
* overwrite salt-formulas with our patched versions;
* create missing links for salt-formulas-* packages (e.g. gnocchi)
which are not available as git repos upstream;
- add mcp.rsa.pub to 'ubuntu' user's authorized_keys for SSH login
to master node to remain backwards compatible;
- mount configuration files for semi-persistent state preservation
at container destroy/rebuild (using '-f' deploy argument);
JIRA: FUEL-383
Change-Id: Ie17e578a7ebeb071b1c0e0a49ad58ffa6bbc89bb
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
|
|
Upcoming docker integration requires special care for network
interfaces Docker will hook to, so virsh networks can be mixed with
Docker networks.
We'll use veth pair(s) to hook together the Docker network(s) to their
counterpart virsh-managed Linux bridge(s).
JIRA: FUEL-383
Change-Id: I1a714dc8be73e79539d85bacac311d1ed05cc18d
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
|
|
Older libvirt (1.x, 2.x, even older 3.x) try to enforce their own MTU
(1500), overriding it back to 1500 after the udev rule finished.
Delay link modification with 1 second and extend its patten to also
include libvirt's own tap interfaces (<network name>-nic).
Change-Id: I3050c11de7f376df02cbf203e9106b61c560683f
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
|
|
Older libvirt seems to raise 'change' events instead of 'add',
so the MTU bump of the libvirt managed tap interfaces was not
properly applied.
Change-Id: Ie1a4f1c4f235ad4f83bf9ee218a40dcedc1ce914
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
|