Age | Commit message (Collapse) | Author | Files | Lines |
|
Add osh into the idf description so that we can deploy osh in baremetal
environments
deploy-scenario:k8-calico-nofeature
installer-type:osh
Change-Id: Id0830ac4744122d149736b1d515af8a1dcbe9eb2
Signed-off-by: Manuel Buil <mbuil@suse.com>
|
|
Currently, all traffic is going through the untagged interface. However,
we are creating interfaces for the different traffic. This patch moves
the private traffic (VM-VM), which normally uses vxlan tunnels, to the
interface created for that
deploy-scenario:k8-calico-nofeature
installer-type:osh
Change-Id: I680698d0bf7ab8045d4ca8f7b5bea8c1e1881d71
Signed-off-by: Manuel Buil <mbuil@suse.com>
|
|
Both nodes are openstack-helm-node-class=primary. That creates problems
when deploying neutron because two l3 agents and dhcp agents are
deployed which sometimes results in problems
deploy-scenario:k8-calico-nofeature
installer-type:osh
Change-Id: Id13ee3cf308f5120061176302573739d7c5a5ed3
Signed-off-by: Manuel Buil <mbuil@suse.com>
|
|
In noha mode, node1 and node2 is used for controller and compute roles.
This patch changes that making node1 only controller and node2 only
compute.
It also modifies a bit the syntax to make it more ansible friendly
deploy-scenario:k8-calico-nofeature
installer-type:osh
Change-Id: I3f40f59ea3c167914ea4c810f7dda60c75fb7175
Signed-off-by: Manuel Buil <mbuil@suse.com>
|
|
JIRA:XCI-11
noha was not passing the tempest test, however, nobody is testing that
test upstream. We should use the test that it being really tested
upstream
deploy-scenario:k8-calico-nofeature
installer-type:osh
xci-flavor:noha
Change-Id: I551e966e81a93120e96027b32096ec18b21bd3bd
Signed-off-by: Manuel Buil <mbuil@suse.com>
|
|
|
|
When deploying manually after a previous deployment, keys and fingerprints of the VMs clash with the stored ones.
This patch clears the relevant files and also fixes an issue with keepalived role which failes ansible lint when already cached.
deploy-scenario:k8-calico-nofeature
installer-type:osh
Signed-off-by: nikoskarandreas <nick@intracom-telecom.com>
Change-Id: I9331a5b7d0ac66a35064dfe7f62dd78a461e570c
|
|
|
|
|
|
|
|
The latest patches in xci provided support for the basic scenario
os-nosdn-nofeature when using the osh installer
deploy-scenario:k8-calico-nofeature
installer-type:osh
Change-Id: Ia37718ae73b91da223bd26715208bb3975a15998
Signed-off-by: Manuel Buil <mbuil@suse.com>
|
|
We should use the Ansible variables whenever possible instead of
hardcoding values like IPs
deploy-scenario:k8-calico-nofeature
installer-type:osh
Change-Id: Iee5f179b0b36d2fcbf9b2eca82a3c31cefd83229
Signed-off-by: Manuel Buil <mbuil@suse.com>
|
|
Ansible best practices recommend doing the definition of variables in
the vars directory instead of directly in the tasks
deploy-scenario:k8-calico-nofeature
installer-type:osh
Change-Id: Ia5e215f8492a901b8c75ac06ee23258249f7f09a
Signed-off-by: Manuel Buil <mbuil@suse.com>
|
|
deploy-scenario:k8-calico-nofeature
installer-type:osh
Change-Id: I0423ab27186316d40c93cce847d7cb9fa7308df1
Signed-off-by: Manuel Buil <mbuil@suse.com>
|
|
deploy-scenario:k8-calico-nofeature
installer-type:osh
Change-Id: Ie198ee7e25579f1438561f71fd965518b067fe42
Signed-off-by: Nikos Mimigiannis <nmimi@intracom-telecom.com>
|
|
This patch creates a new installer tree in xci that uses
openstack-helm to deploy openstack on a kubernetes cluster.
USAGE: Export INSTALLER_TYPE=osh, DEPLOY_SCENARIO=k8-calico-nofeature
and XCI_FLAVOR=noha or mini and run xci-deploy.sh as in documentation.
deploy-scenario:k8-calico-nofeature
installer-type:osh
Change-Id: I212f70eb51c2a38c798c11367d2ebb8bf5f4a1de
Signed-off-by: nikoskarandreas <nick@intracom-telecom.com>
|
|
Change-Id: Id0b7becd47534de8b80df93e85daf399bc613aec
Signed-off-by: Manuel Buil <mbuil@suse.com>
|
|
We should try to use pip ansible module as much as possible. If we use
pip in the command line we run into issues when moving towards a newer
version of bifrost
Change-Id: Ib606e0cc0e754f0a9ca07bf7724b7f8d70976b2d
Signed-off-by: Manuel Buil <mbuil@suse.com>
|
|
The ara module just released an update
which changes how the module can be imported.
As a result, the xci-deploy script immediately
fails when trying to setup the virtualenv.
This patch sets ara to its previous
version (ara==0.16.4).
Change-Id: Ic54d18dc877857389e8b36bbf2e8bf4b70bc2812
Signed-off-by: Nikos Mimigiannis <nmimi@intracom-telecom.com>
|
|
lfpod4 uses pod4-nodeN notation and that was missing for the kubespray
deployer
Change-Id: If5b0a600b65f830c96ff8888986e0ef8decaada6
Signed-off-by: Manuel Buil <mbuil@suse.com>
|
|
|
|
Change-Id: I6086d13450f95a5c901270c7d4b69a0aa127f129
Signed-off-by: Manuel Buil <mbuil@suse.com>
|
|
When working with an interface that uses vlan, the vlan is created when
executing the ifup command. As we previously execute the command:
/sbin/ip addr flush dev {{ item }}
it will not return code 0 because the interface does not exist yet.
Consequently, the second part of the command (ifup) is never executed
and the new network config is never triggered.
This patch removes the requirement to get a rc=0 in the first command to
run the second one.
Note that /sbin/ip addr flush dev {{ item }} must be run before ifup
when working with non-vlan interfaces that already have an ip, otherwise
ifup fails in Ubuntu
Change-Id: Ieac273a49b78104f24d4c1ba81c263b450a40a9f
Signed-off-by: Manuel Buil <mbuil@suse.com>
|
|
The dns keys should only appear in case the idf provides a dns entry for
the network. Otherwise, it should not exist at all. If it exists, even
if the value is empty, ansible will transform it in a variable and
"item.network.dns is defined"
will return true:
https://github.com/opnfv/releng-xci/blob/master/xci/playbooks/roles/bootstrap-host/templates/osa/debian.interface.j2#L35
A docstring is also added to explain what is the purpose of the class
Signed-off-by: Manuel Buil <mbuil@suse.com>
Change-Id: Ib8afa06cecb54f384083060073fa463c7f8d313f
|
|
deploy-scenario:os-nosdn-nofeature
installer-type:osa
Change-Id: Icebfbbd151c53b9ca3f38c7fdb107cb1f51b2b3e
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
Baremetal is failing because ironic takes long to transfer the image to the
hard drive of the nodes
Change-Id: Ief704e92307d1ea7fe55ee0268abae49e0126503
Signed-off-by: Manuel Buil <mbuil@suse.com>
|
|
Change-Id: I3977c3d99b06b2548259e8765c262eb996ba30cf
Signed-off-by: Manuel Buil <mbuil@suse.com>
|
|
The name of the file should have baremetal not barematal
Change-Id: I15d70b69943e8ce3032c76d1cd7bc7272a6b8d56
Signed-off-by: Manuel Buil <mbuil@suse.com>
|
|
Moving to the newer SHAs of stable/rocky
installer-type:osa
Change-Id: I89de6554d5e3bef8b2b49c6a3e621d3ca3a6f4dc
Signed-off-by: Manuel Buil <mbuil@suse.com>
|
|
To read the idf when the installer is kubespray, we need the network
details too
Change-Id: Idb9b0a4338a224e146abc78690067659bc94c302
Signed-off-by: Manuel Buil <mbuil@suse.com>
|
|
The PDF/IDF filenames to use during deployments in CI will be generated
dynamically based on which slave the job is running on with the help of
the SLAVE_NAME environment variable Jenkins injects into job environment.
It will probably look like this
pdf=var/pdf.yml
idf=var/idf.yml
if [[ "$SLAVE_NAME" !~ virtual ]]; then
pdf=var/${SLAVE_NAME}-pdf.yml
pdf=var/${SLAVE_NAME}-idf.yml
fi
./xci-deploy.sh -i $idf -p $pdf
deploy-scenario:os-nosdn-nofeature
installer-type:osa
Change-Id: Ief319ee36292ca888b97e4059a26337ee98dfef2
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
The IDF files contain DNS information so we should respect that when
we configure the various XCI nodes. The DNS information is also a
list instead of a string so treat it as such.
Change-Id: I1c4d5eb600baaca35b2838dcafa7a75e59bf6783
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
|
|
|
|
|
|
|
|
We are deploying the NFS server in computes. In all flavors, node2
takes the role of compute00 and thus the NFS server is in
172.29.244.12. Therefore, the openstack config for ha is wrong
Change-Id: I5e82ddd670b44e291c0b866ba4fde57e74b68643
Signed-off-by: Manuel Buil <mbuil@suse.com>
|
|
When deploying baremetal, the traffic should be segmented using the
different interfaces instead of through vlans. All the config is done
based on idf and pdf information.
When doing non-baremetal, opnfv and nodes get the same config. When
doing baremetal, opnfv and nodes get a different network config
Change-Id: I23aa576bc782c7c69d511a5558827110c37b558a
Signed-off-by: Manuel Buil <mbuil@suse.com>
|
|
When deploying baremetal, the traffic should be segmented using the
different interfaces instead of through vlans. All the config is done
based on idf and pdf information.
When doing non-baremetal, opnfv and nodes get the same config.
When doing baremetal, opnfv and nodes get a different network config
Apart from that, if vlan_id is defined in the name, there is no need
for VLAN_ID in the interface descriptor. This simplifies things
Change-Id: Iddbb90af807b43e247e5ee11fe735df9e823d4bf
Signed-off-by: Manuel Buil <mbuil@suse.com>
|
|
|
|
Bifrost is exposing quite a few variables so we need to collect them
as well.
Change-Id: I7e7ca7a093f35a0acb53af360e58444f6c1de7e4
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
The current configuration only dumps the booting info into a file and
the pty does not work (i.e. virsh console opnfv returns a failure
because it cannot find a character device). After some investigations,
it is apparently impossible to have both active:
https://github.com/Mirantis/virtlet/issues/249
Therefore, we should remove the pty part of the xml. To connect to the
VM in case of network problems, we can always use vnc.
Apart from that, the console part is not necessary as libvirt will
create that one for us
Change-Id: I80a59163b4ba4e6bff34cb5378893201e93ddb87
Signed-off-by: Manuel Buil <mbuil@suse.com>
|
|
installer-type:kubespray
deploy-scenario:k8-nosdn-nofeature
Change-Id: I6b59df5112e9b3459bf3147557f5f22fe0fb778b
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
Change-Id: I50c367433dc8cf8964c291c916ea939e25f638cb
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
When we install packages from the distro repos we need to make sure
that the database is updated. This also takes SUSE's zypper pkg manager
into consideration which can benefit from the same Ansible option.
Change-Id: I7a2206bfc5827b9ccb448278759711c560bb4679
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
Currently we were taking the names of the nodes to be generated from the
NODE_NAMES variable whereas ansible was fetching the names from the
dynamic_inventory.py which uses idf. This resulted in problems: when
doing ha, ansible was provisioning a compute as a controller and
vicebersa.
This patch forces create_nodes role to fetch the name from idf and thus
align with the naming schema of ansible
deploy-scenario:k8-calico-nofeature
installer-type:kubespray
Change-Id: Id1473727405701fd9ed0cb2f1394ee8676cec337
Signed-off-by: Manuel Buil <mbuil@suse.com>
|
|
Right now, we only support ipmi hosts (either virtual or physical) and
that is why our json is always describing the ipmi parameters. It does
not make sense that we have a variable which would allow to change that
Change-Id: I7b88aca5930a73d68342e3d4cf21f9e96286c4d7
Signed-off-by: Manuel Buil <mbuil@suse.com>
|
|
The variable should be host_group which is generated at line 32 and 35
Change-Id: I7add3af73198ec0638dee0c8f189a3a372a78ee8
Signed-off-by: Manuel Buil <mbuil@suse.com>
|
|
Since the patch:
https://gerrit.opnfv.org/gerrit/#/c/63173/
The variable BIFROST_USE_PREBUILT_IMAGES has changed to
BIFROST_CREATE_IMAGE_VIA_DIB. As jenkins does not trigger testing jobs
when editing file in xci/scripts/, this change is done in a separate
patch
Change-Id: I3bc285936fae5b7514272ca0ad2418b60446e4aa
Signed-off-by: Manuel Buil <mbuil@suse.com>
|
|
|