Age | Commit message (Collapse) | Author | Files | Lines |
|
- submodule: bump Pharos to pick up installer adapter support, as well
as IDF updates for Ericsson baremetal pod1;
- labs/local/virtual: Bump mgmt, public networks start addresses from
.1 to .10, similar to ericsson-pod1;
While at it, drop patch now upstream and instead adopt the new param
'neutron:server:root_helper_daemon'.
JIRA: FUEL-351
Change-Id: I9bc244a7fd8698861a390ed2b6a27804be46c285
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
|
|
* class including order (default was used)
* disable root_helper_daemon (incompatible with UCA)
* turn off websocket for pseudo agent updates
Change-Id: I4d7971d393da184bdd55f65b4d3fd8d9e898543f
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
|
|
Change-Id: If7d51555bc13dbcaa63f93ab1993f3655e2ce643
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
|
|
Change-Id: I503c8ad32900406e1fa375cec9a91b454889d8bf
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
|
|
Change-Id: Icee56da3720f0926e42390965581639f6a344b77
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
|
|
Now we explicitly add a LVM PV on /dev/sda{1,2} for Cinder storage,
so we can safely drop the patch disabling LVM volume filtering.
If we later move the PV to a different disk, we can just add the VG
and LV definitions to linux:storage via reclass, and Salt will skip
setting them up (as they're already created by MaaS), yet keep the
filtering sane.
While at it, fix 'nova_cpu_pinning' param expr; constructs based on
reclass interpolation (e.g. '${_param:x}') do not work when
parameters are passed via reclass.storage templating, so change
reclass interpolation syntax with classic YAML anchors.
Fixes: 672ae12
Change-Id: Ieb41635ddeb630543d7e4d1079f45d636d9a43af
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
|
|
- fix `route-br-ex` if-up.d script failing when route already exists
by adding a wrapper around distro's '/sbin/route' binary in
'/usr/local/sbin/route', exploiting default order in Ubuntu PATH;
- fix 'br-prv' duplicate entry in 'interfaces.d/ifcfg-br-prv' and
'interfaces' caused by upstream bug [1];
- add barrier waiting for all baremetal nodes online before attempting
reboot, trying to catch rare failures which are undetectable in logs
as both a succesful reboot and a disconneted minion report 'n/c';
With the above in place, networking service should no longer fail
to start on cmp nodes w/ DPDK.
[1] https://github.com/saltstack/salt/issues/40262
Change-Id: I6d4895376ce323c14c997e6c9af2ea3eeeee0184
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
|
|
Drop one questionable patch responsible for MaaS node authorized
keys to include mcp.rsa.pub by reading the contents of authorized
keys on mas01, assuming mcp.rsa.pub will be on the first line.
Instead, export the contents of the public key using a shell env
var during deploy, which gets expanded via maas_pdf j2 template
into a reclass param, leveraging existing salt-formula-maas sshprefs
mechanism for delivering the key to MaaS.
Since we require the public key to exist before expanding templates,
move `generate_ssh_key` call outside the current infrastructure
handling block, allowing it to execute during all `deploy.sh` calls,
even for dry-runs.
Change-Id: I0f53b0f764a2fafd292e0ffd399c284acf61bd30
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
|
|
Add a new class of scenarios, based on existing baremetal HA
scenarios, but instead of having a virtualized control plane (VCP),
all Openstack controller services will run directly on the cluster
nodes.
This change adds the common scaffolding, as well as the OVS scenario.
The new scenario(s) can be used on full-baremetal clusters, soon on
full-virtual clusters and later on hybrid (virt + bare) clusters.
This change defines old (current) style scenario definitions for
both baremetal and virtual, both named:
- os-nosdn-nofeature-novcp-ha;
Prerequisites:
1. Merge-able by name reclass.storage.node definitions
Each cluster (e.g. database, telemetry) adds its own set of
reclass storage node defitions, which for novcp scenarios should
be merged into a single node (kvm) based on the 'name' property.
This is not currently supported by upstream reclass 'node.sls'
high state, so add support for it via an early patch (required
before salt-master-init.sh tries to handle reclass.storage).
2. common reclass classes for novcp
Some of the classes in `baremetal-...-common-ha` are not fit for
novcp as they define VCP-specific config/inheritance, so add new
versions of said classes with novcp in mind or adapt old classes:
- parameterize ctl hostname in `openstack_compute.yml`;
- new `openstack_control_novcp.yml`;
- new `openstack_init_novcp.yml`;
3. Handle hard set names in state files for baremetal nodes
Some of our state files (e.g. maas) hardcodes baremetal node names
to 'kvm', 'cmp', so we need to align the names in novcp scenario
with these values to re-use the maas state. As a future improvement
we should parameterize these names in all state files.
As a consequence, our baremetal controller nodes will also use
'kvm*' hostnames (instead of 'ctl*').
4. Add 'noifupdown' to all interfaces on kvm nodes to prevent duplicate
IPs/routes created at *any* ifup due to /etc/network/route-br-ex.
Patch salt-formula-linux to skip network restart on 'noifupdown',
also when routes are present on that interface.
JIRA: FUEL-310
Change-Id: Ic67778f63e5ee0334dbfe9547c7109ec1a938d61
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
|
|
Based on Canonical research (https://goo.gl/QJykMa) there is
low-risk of attack for private clouds environments, therefore
turn off the related kernel patches & regain performance back.
Change-Id: I661fa127241e327b07d21a29d58d584997607123
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
|
|
* switch ovs/dpdk scenario from vlan to vxlan mode
* force br-ex interface to mitigate race with incorrect state
* remove dpdk packages list (already in upstream)
Change-Id: Ib827cef2d67879fd2a86d286ca2118b22493274d
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
|
|
Instead of defining a http proxy for all salt-minion traffic, which
also includes some Openstack API accesses we can't filter (no_proxy
is not yet supported), add & leverage support for proxy configuration
during APT keyserver access / key download.
JIRA: FUEL-331
Change-Id: I9470807633596c610cfafb141b139ddda2ff096b
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
|
|
Since Mirantis prebuilt image comes with salt-minion 2016.3 instead
of 2016.11 and upgrading it leads to a hard to break catch-22, use
the Ubuntu cloud archive image we already download for FN VMs and
pre-install:
- a newer kernel (hwe-edge);
- salt-minion (2016.11);
This also implicitly aligns the image handling on AArch64 and x86_64.
Change-Id: I86d1c777449d37bdd0348936a598e3ffe9d265af
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
|
|
Unlike nightly dist, stable/2017.12 distributions of salt formulas
repo do not yet include this change, so bring it back.
This reverts commit 8fbafdf8a665fb8fff4d6f9f14c343e109c122ec.
Change-Id: I7f7011750d385a28f4653faeeb74edb1cac1bcf2
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
|
|
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>
|
|
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>
|
|
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>
|
|
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>
|
|
- 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>
|
|
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>
|
|
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>
|
|
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>
|
|
The patch has been merged to upstream linux formula.
Change-Id: I09f96be920c5c3ecd09fd2ab132e25519902239b
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
|
|
The upstream commit has been merged
https://github.com/salt-formulas/salt-formula-maas/commit/0c622327
This reverts commit 2c48915a1a77aa3ac0133155fd0cdadbf280493d.
Change-Id: Iff86f3dc171371903e3294eed43977eb6197a5e7
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
|
|
Change-Id: I603b39d46d3f2f2448b8f9746ad0745e416bbee7
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
|
|
JIRA: FUEL-304
Change-Id: I52ae89f933f976b678b27f79e2ad67c3699ef8ec
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
|
|
Previous commit pinned the salt-formula-package by installing a
specific version after OPNFV-specific patches were applied by
`patch.sh`. To prevent patch bypass, move the pin to reclass model.
While at it, also move salt-formula-neutron pinning to reclass to
bypass wrong minimum length of 2 for version-pinned formula pkgs.
Change-Id: I35c2773c130d79ec919685cd771311d6c11e11a8
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
|
|
Horizon service definition hardcodes the API version to v3, and
also overrides the value in our reclass model during interpolation.
Adjust the default to '2'.
While at it, move 'reclass' patching later in `salt.sh`, after
the salt formula packages are installed.
JIRA: FUEL-284
Change-Id: I6f63b4d00ae85c82b076b3efef857e7cc4f3fd35
Signed-off-by: Ting Wu <ting.wu@enea.com>
Signed-off-by: Junaid Ali <junaidali.yahya@gmail.com>
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
|
|
While at it, compact 'set' into bash shebang where possible and
add `make patches-copyright` target to simplify adding patch
license headers.
Change-Id: I0c841de72e5709e5eef915a52c5ec4a7fc0f7c37
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
|
|
Change-Id: I8dba3676adc8cc49731d91db7cc028a9c5b1627d
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
|
|
- add new patch for maas.region, extending it poorly with a timeout
override mechanism; the new comissioning/deploying timeout defaults
(10/15min) will be used instead of MaaS defaults (20/40min), unless
reclass params are defined with different values;
- add 30s delay between 'machine mark-broken' and 'machine mark-fixed'
MaaS cli commands (fixes a rare race condition);
- fix forgotten replace in 'maas.pxe_route': s/opnfv_fuel_/opnfv_/g;
Change-Id: I71c562b80031bac2793dd470d52928c2d62e5300
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
|
|
JIRA: FUEL-282
Change-Id: I6c86ce0b1113ca674b1756e7997559eee90a4e5f
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
|
|
JIRA: FUEL-274
Change-Id: I2c8161b24cb18a0d1f9dc6fd509ce18af7ea8cf5
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
|
|
Previously, we hardcoded the fabric name for our 3rd interface
(which serves PXE/DHCP for the target nodes) to "fabric-2",
relying on predictable index numbers to be provided by MaaS based
on the interfaces defined in /etc/network/interfaces.
However, the fabric IDs/names generated by MaaS are not predictable,
and therefore cannot be hardcoded in our reclass model / scripts.
Work around this by:
- adding support for fabric ID deduction based on CIDR matching
during subnet create/update operation in MaaS py module;
- adding support for VLAN DHCP enablement to MaaS py module,
which was previously handled via shell MaaS API operations
from maas/region.sls;
While at it, revert previous commit that disabled network discovery
("MaaS: Disable network discovery"), since it turns out that network
discovery was not the culprit for subnet creation failure, but wrong
fabric numbering.
This reverts commit 8cdf22d1a1bae4694a373873cab4feb6251069b7.
Change-Id: I15fa059004356cb4aaabb38999ea378dd3c0e0bb
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
|
|
- ci/deploy.sh: fail if default scenario file is missing;
- start by copying reclass/classes/cluster/virtual-mcp-ocata-ovs as
classes/cluster/baremetal-mcp-ocata-ovs;
- add new state (maas) that will handle MaaS configuration;
- Split PXE network in two for baremetal:
* rename old "pxe" virtual network to "mcpcontrol", make it
non-configurable and identical for baremetal/virtual deploys;
* new "pxebr" bridge is dedicated for MaaS fabric network, which
comes with its own DHCP, TFTP etc.;
- Drop hardcoded PXE gateway & static IP for MaaS node, since
"mcpcontrol" remains a NAT-ed virtual network, with its own DHCP;
- Keep internet access available on first interfaces for cfg01/mas01;
- Align MaaS IP addrs (all x.y.z.3), add public IP for easy debug
via MaaS dashboard;
- Add static IP in new network segment (192.168.11.3/24) on MaaS
node's PXE interface;
- Set MaaS PXE interface MTU 1500 (weird network errors with jumbo);
- MaaS node: Add NAT iptables traffic forward from "mcpcontrol" to
"pxebr" interfaces;
- MaaS: Add harcoded lf-pod2 machine info (fixed identation in v6);
- Switch our targeted scenario to HA;
* scenario: s/os-nosdn-nofeature-noha/os-nosdn-nofeature-ha/
- maas region: Use mcp.rsa.pub from ~ubuntu/.ssh/authorized_keys;
- add route for 192.168.11.0/24 via mas01 on cfg01;
- fix race condition on kvm nodes network setup:
* add "noifupdown" support in salt formula for linux.network;
* keep primary eth/br-mgmt unconfigured till reboot;
TODO:
- Read all this info from PDF (Pod Descriptor File) later;
- investigate leftover references to eno2, eth3;
- add public network interfaces config, IPs;
- improve wait conditions for MaaS commision/deploy;
- report upstream breakage in system.single;
Change-Id: Ie8dd584b140991d2bd992acdfe47f5644bf51409
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
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>
|
|
Workaround the issues described in [1], stating that salt-formula-maas
package cannot set an autogenerated PostgreSQL password, respectively
the known error thrown during initial setup.
FIXME: These should be reverted later, after fixing the MaaS password
update and initial artifact sync in the salt formula / scripts.
[1] https://docs.mirantis.com/mcp/1.0/mcp-deployment-guide/\
install-base-infra/set-up-bare-metal-provisioner/configure-maas-vm.html
Change-Id: I8b37f55d3caa4119c64f9549578850dd4eb9e3ad
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
|
|
salt.sh currently clones the full Fuel@OPNFV git repo from upstream
public mirror, preventing us from testing locally edited or new
patches.
Instead, bring back git submodule handling from old f_repos, clone
and patch each submodule locally, then copy the whole parent repo
over to cfg01.
This is also a first step towards implementing offline deploy support.
NOTE: This adds new deploy prerequisite packages:
- git (for submodule clone/update);
- make (for submodule patching);
- rsync (for parent repo replication to cfg01);
NOTE: Parent repository is expected to be a git repo, in order to
work with git submodules.
While at it, perform some minor related changes:
- add deploy artifacts (ISOs, qcow2 files) to .gitignore, also used
to filter-out such files during rsync to cfg01;
- remove obsolete Fuel patches (old f_repos mechanism);
- rename "reclass-system-salt-model" submodule;
Change-Id: I6210d80d41010b2802e4f1b31acf249a18db7963
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
|
|
Change-Id: I7bb984880e98b7bdec9aa0b3895a3be9fd75cac0
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
|
|
For opendaylight support:
* neutron formula
* reclass system model
Plus missing apply of haproxy state.
Change-Id: Ic9e3672b51d5331656d636c44c3f891e49437e23
Signed-off-by: Michael Polenchuk <mpolenchuk@mirantis.com>
|