Age | Commit message (Collapse) | Author | Files | Lines |
|
Change-Id: Ie2d473d3cdc804f698244f163c62bed30a5dd33d
Signed-off-by: wutianwei <wutianwei1@huawei.com>
(cherry picked from commit 3e8cc46ac37c5b89b606bd2ee1e3b389c84fd7fa)
|
|
JIRA: COMPASS-595
set enable_proxy_headers_parsing to True in neutron config file
Change-Id: I63a0795cad02051164b0881b8b33ac14a1562c9d
Signed-off-by: wutianwei <wutianwei1@huawei.com>
|
|
Change-Id: I89b15e7e200d37ac5d44cfee974f0f1c63981625
Signed-off-by: wutianwei <wutianwei1@huawei.com>
|
|
Change-Id: Ieb5d505adc88644cfc1f802364064f5551a1ed53
Signed-off-by: wutianwei <wutianwei1@huawei.com>
(cherry picked from commit 01ff42530757cf99e99b0d63d6daa822b93e1941)
|
|
JIRA: COMPASS-593
OpenStack Ansible install latest pip on
each contoller node and the latest pip
(currently 10.0.0b1) doesn't support empty
package list to be passed into it. Add a
user config file to fix pip version.
Change-Id: Ie193389c2b4995cdc4b81876f8cdcfc512cf5ea1
Signed-off-by: Harry Huang <huangxiangyu5@huawei.com>
(cherry picked from commit 18cf3eec4990e2e747cadc8b9c177e5b1d0c67f4)
|
|
Change-Id: I4e7b47a8f2021a1cf25b80d7818f4dabb441c39f
Signed-off-by: wutianwei <wutianwei1@huawei.com>
(cherry picked from commit 4ce9351a4802a1bdaecdf1a0fe982e2878755c03)
|
|
|
|
1. using horizon group limit
2. set fact openstack_release to hostvars of
localhost
Change-Id: I0af0da1a508e862d0e2bb975f29897ec2948617f
Signed-off-by: Harry Huang <huangxiangyu5@huawei.com>
|
|
JIRA: -
By default, os_tacker will use "{{ tacker_bin | dirname }}/etc/tacker"
as tacker_etc_dir. This will cause a failure when keystone creates
fernet key for VIM in tacker due to keystone uses "/etc/tacker" by
default. So this patch is proposed to set tacker_etc_dir to
"/etc/tacker".
Change-Id: Icbffaeb1376f83b4c9904b0258139ce1518ac9f8
Signed-off-by: Yifei Xue <xueyifei@huawei.com>
|
|
JIRA: COMPASS-590
From the time being, Functest ODL testcase calculates ODL ip thanks to Neutron IP.
It should be calculated on Installer side (releng jobs) and set via SDN_CONTROLLER_IP in Functest env instead.
https://git.opnfv.org/functest/tree/functest/opnfv_tests/sdn/odl/odl.py#n185
It ensures that Functest can rely on SDN_CONTROLLER_IP in all cases (OpenStack end users included)
Change-Id: I01c0af952dbb21592aa973daaea3845346dc833f
Signed-off-by: Yifei Xue <xueyifei@huawei.com>
|
|
|
|
Support deploying multus sriov CNI plugins by setting
environment "kube_network_plugin" to "sriov".
Change-Id: I3672fd7b6036063bdee57450c2100f39aa5ef68b
Signed-off-by: Di Xu <di.xu@arm.com>
|
|
JIRA: -
Ks8 deployment doesn't install git, this will cause failure in
Stor4nfv scenarios. So proposing this patch to solve this.
Change-Id: I98ab0b3a2a5f46eb1eebd4ea94fb60ccafe10272
Signed-off-by: Yifei Xue <xueyifei@huawei.com>
|
|
If don't setup network, the host cannot connect to the Internet on baremetal pod.
So we should update cache after setting the network.
Change-Id: Ief0eb2d82c67b0ace2cfb233b9792f6e009a1524
Signed-off-by: wutianwei <wutianwei1@huawei.com>
|
|
Don't run storage roles when scenario isn't stor4nfv
Change-Id: Ia8e65c90b58533427e74bad225aaa962d2583934
Signed-off-by: wutianwei <wutianwei1@huawei.com>
|
|
JIRA: -
Add some configurations to enable CSI plugin in k8s v1.9 which
is used by Stor4nfv scenario.
Change-Id: I78e0b39f31ff666eace50921c691107154dcb249
Signed-off-by: Yifei Xue <xueyifei@huawei.com>
|
|
- Adds 'stor4nfv' plugin which installs OpenSDS using stor4nfv.
- Adds k8s-nosdn-stor4nfv-ha scenario file.
- Adds k8s-nosdn-stor4nfv-noha scenario file.
Change-Id: Iee718a48994ad4e55f4b7a01933bd4e7773ac67f
Signed-off-by: ramamani yeleswarapu <ramamani.yeleswarapu@intel.com>
|
|
Support deploying multus multiple flannel CNI plugins by setting
environment "kube_network_plugin" to "2flannel".
Change-Id: I23a3d42452b4a5e0858a18934f508649f5961e7a
Signed-off-by: Di Xu <di.xu@arm.com>
|
|
|
|
JIRA: -
1. compass4nfv can deploy v1.9.1 kubernetes
2. end user can use the kubernetes dashboard
Change-Id: I274dda38599da95f382dfcb4159b530da8422fd8
Signed-off-by: hu xinhui <xinhui_hu@foxmail.com>
|
|
JIRA: COMPASS-573
1. With HORIZON_IMAGES_UPLOAD_MODE set to direct
Horizon provides the endpoint for Glance based on
OPENSTACK_ENDPOINT_TYPE. If OPENSTACK_ENDPOINT_TYPE
is set to internalURL any browser outside the internal
network is unable to upload image. Add ansible task to
set HORIZON_IMAGES_UPLOAD_MODE to legacy as a workaround.
2. Add ansible lookup plugin to get openstack release
3. set openstack_release into group_vars/all in config-osa
to make this variable readable for other tasks
Change-Id: I9ef358e1f4acb0c329a032e18359de12284f3b56
Signed-off-by: Harry Huang <huangxiangyu5@huawei.com>
|
|
JIRA: -
Currently there is no variable definition file of tacker
for centos in os_tacker repo. This will lead an error
while including vars on centos deployment. So this patch will
add a file named "redhat-7.yml" for centos.
Change-Id: I965834adb0d2d25e5a68a9c7dbb397342108078f
Signed-off-by: Yifei Xue <xueyifei@huawei.com>
|
|
|
|
Adds/updates the following to install influxdb and grafana on the
controller. Configures them to receive collectd data from the compute.
/plugins/barometer/roles/influxdb
/pluigns/barometer/roles/grafana
/deploy/adapters/ansible/openstack/HA-ansible-multinodes.yml
Change-Id: Ie33489e9834da5bb5c77b392451b81cf44a849d6
Signed-off-by: ramamani yeleswarapu <ramamani.yeleswarapu@intel.com>
|
|
|
|
JIRA: COMPASS-579
the script collect-log.sh would be called by releng,
after all the deploy and tests job are complete.
Change-Id: Ic2604ca0650989bc13310377427694ededf75a9d
Signed-off-by: wutianwei <wutianwei1@huawei.com>
|
|
JIRA: COMPASS-577
1. Add tacker service into haproxy configuration file
2. Include tacker installation into setup-openstack.yml
Change-Id: I5a91766e034ec55af711029aac4ba9ed4a388db2
Signed-off-by: Yifei Xue <xueyifei@huawei.com>
|
|
|
|
1. Change the gnocchi storage to ceph.
2. Change the size of loopback to 500GB
3. Change the ceilometer polling yaml
4. Run ceilometer-upgrade command
and restart ceilometer service
Change-Id: I8948c7b7e340c11e5d05c06a7d54c0e1b807620c
Signed-off-by: wutianwei <wutianwei1@huawei.com>
|
|
JIRA: COMPASS-571
The network trunk service allows multiple networks to be connected to an
instance using a single virtual NIC (vNIC). Multiple networks can be presented
to an instance by connecting it to a single port. This is a highly relevant
feature for VNFs and hence for OPNFV as a reference NFVI.
This patch is about enabling the trunk port feature in Compass deployments.
Change-Id: I270c3f1d5e4af99578c8e1882fae12ddddaabe78
Signed-off-by: Yifei Xue <xueyifei@huawei.com>
|
|
|
|
kubespray for default, but k8s apiserver HA is not implemented by kubespray, This patch aim is to achieve the k8s apiserver HA Change-Id: I805b5eb2f4efa7ca82fcef7bfd3f4cad35ed65b5 JIRA: - Signed-off-by: hu xinhui <xinhui_hu@foxmail.com>"
|
|
|
|
|
|
compass installer deploy k8s using kubespray for default,
but k8s apiserver HA is not implemented by kubespray, This
patch aim is to achieve the k8s apiserver HA
Change-Id: I805b5eb2f4efa7ca82fcef7bfd3f4cad35ed65b5
JIRA: -
Signed-off-by: hu xinhui <xinhui_hu@foxmail.com>
|
|
Cluster name now can be set in DHA file
which makes ansible dir changeable. Use
run_dir variable to point to correct dir.
Change-Id: I673b0b472b86ce505a8ea29ce4fb3666b4b991bd
Signed-off-by: Harry Huang <huangxiangyu5@huawei.com>
|
|
JIRA: COMPASS-575
Add missing condition check for subnet create
task.
Change-Id: Ibd25a7e7e4edd40be6b0f14bb8bffb828a65f648
Signed-off-by: Harry Huang <huangxiangyu5@huawei.com>
|
|
from kube_master fetch the admin.conf
to the directory /opt/admin.conf of compass-tasks docker
Change-Id: I1c236414c968f721d04fb6da0dc7a9514c892f93
Signed-off-by: wutianwei <wutianwei1@huawei.com>
|
|
JIRA: -
Use the ntp server in compass-cobbler.
Change-Id: Ia5d3d8fb32ce51667841cec74622c7f0da929587
Signed-off-by: Yifei Xue <xueyifei@huawei.com>
|
|
JIRA: -
Open log server of OpenStack Ansible to collect all of the
logs in each LXC. Then it will be very convenient to debug.
Change-Id: Iafd625a2af4eea04ff7528c9d8007bed1b218822
Signed-off-by: Yifei Xue <xueyifei@huawei.com>
|
|
Due to OSA commit id rollback, open this fix_pip_version
to fix ldap issue
Change-Id: I9d41f0dcb04ecdc3d73b92a5da6fa254b5246045
Signed-off-by: Harry Huang <huangxiangyu5@huawei.com>
|
|
|
|
1. Don't check failed=1 instead of checking the last task name.
2. comment the fix pip version yml
Change-Id: I3c22074f319a1d0f1ae0f3df43ca6e10ec684aac
Signed-off-by: wutianwei <wutianwei1@huawei.com>
|
|
|
|
1. install pkgconfig
2. mv the tasks of fixing pip issue out of the main script
Change-Id: Icc08a3c461690d1d6747455479df0a58d87d22da
Signed-off-by: wutianwei <wutianwei1@huawei.com>
|
|
Currently, tcp_tw_recycle is set explicitly in sysctl.conf when
installing OS by cobbler. This option is removed in kernel4.12,
current sysctl.conf breaks sysctl function and K8s deployment.
Actually, tcp_tw_recycle is considered broken and not recommended to
use. We may simply drop this option to be compatible with different
kernel versions.
References:
- https://vincent.bernat.im/en/blog/2014-tcp-time-wait-state-linux
- https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=4396e46187ca5070219b81773c4e65088dac50cc
Change-Id: I2cd89db3d423c811ba58eb5960f3cc718475be5f
Signed-off-by: Yibo Cai <yibo.cai@arm.com>
|
|
|
|
|
|
JIRA: -
Set "security_ntp_servers" to use only one server;
Correct the path of chrony key for centos deployment.
Change-Id: I3050018fe07c51912e0b52a1c5c7ce2dd4d3f6ce
Signed-off-by: Yifei Xue <xueyifei@huawei.com>
|
|
|