Age | Commit message (Collapse) | Author | Files | Lines |
|
We need to flush the handle to restart network,
before checking network.
improve the message when failed
Change-Id: If38198c62b4bf81d38b294264db30514998fe086
Signed-off-by: wutianwei <wutianwei1@huawei.com>
|
|
We use openstack-ansible to deploy the openstack,
the position of openrc is different from the before.
change the apt source when use local apt repository
Change-Id: I1c5492b294a8645f452868c9cf2fec2e0802f823
Signed-off-by: wutianwei <wutianwei1@huawei.com>
|
|
JIRA:-
Replace OSA default address with ntp_server address.
OSA will use it to check internet connection.
Rename some roles that make users confused.
Change-Id: I4dd7e242e427e5bc3a611450a3bc436e44a8fefe
Signed-off-by: chigang <chigang@huawei.com>
|
|
JIRA:-
use OpenStack-ansible deploy openstack, so remove obsoleted code.
Some of the enhanced features will be added in later versions
Change-Id: Ie92b92b5de234a7d7d03b578b0bc15fd0218b3b3
Signed-off-by: chigang <chigang@huawei.com>
|
|
JIRA:-
It doesn't need to config DNS in ansible roles, just need to
set NAMESERVERS variable
Change-Id: I37a70b32c766c1fad7afa5289ffbbc27b956b2d5
Signed-off-by: chigang <chigang@huawei.com>
|
|
1. change the apt sorce from http://hk.archive.ubuntu.com
to http://archive.ubuntu.com
2. if setup host failed, destroy the failed lxc. then retry to setup.
3. fix the dns-nameserver bug
Change-Id: Ic609c75b50496e2174ab09425bce45a0ec2c09c0
Signed-off-by: wutianwei <wutianwei1@huawei.com>
Signed-off-by: chigang <chigang@huawei.com>
|
|
1. replace the linuxbridge with openvswitch
2. create the network and flavors
3. fix the haproxy when public endpoint using http
Change-Id: I6fe93d0c4d5305e5dcab8d2e6c628eb5fa23ab56
Signed-off-by: wutianwei <wutianwei1@huawei.com>
|
|
1. install the yum packages
2. add the ifcfg scripts to adapt Centos7
3. change MaxSession to 100
4. if setup-host failed, retry to run the failed container
Change-Id: I57589d38f6af1677cb889875c85d6b8bac5c80d0
Signed-off-by: wutianwei <wutianwei1@huawei.com>
|
|
The network restart handler comes at the end
of all tasks which cause pkg install task be
executed when network is not ready. Flush this
handler right after configure network tasks.
Change-Id: I98b9f6f0f2ec0b79632188a481673529a1b522bb
Signed-off-by: Harry Huang <huangxiangyu5@huawei.com>
|
|
Before network is setup, host can't access to internet.
Install bridge-utils, ifenslave and vlan when cobbler
managing OS installation so when network restart all pkgs
needed is ready.
Change-Id: Idaa880ed0abfb50f417d67fe55e964992a185c80
Signed-off-by: Harry Huang <huangxiangyu5@huawei.com>
|
|
1. delete package part when external network is not ready
2. modify some vm_environment ip to adapt with containerized
compass4nfv code
Change-Id: I9d777675e7ca9d6f9b12a885df8677b2446e6917
Signed-off-by: huangxiangyu <huangxiangyu5@huawei.com>
|
|
JIRA: COMPASS-534
1. rm compass vm and add ansible to bring up 5 compass
containers
2. use tar package instead of compass.iso which contains
compass docker images, OS ISO, PPA, pip packages.
3. modify client.py to communicate with containerized
compass-core
4. modify cobbler files and ansible callback files
to adapt with containerized compass-core
5. upgrade openstack version to ocata
6. use the openstack-ansible to deploy openstack
7. virtual deploy external use nat
Change-Id: Ifa2a3f5b8c7c32224ac4276fd3d4cc2b0d270a26
Signed-off-by: huangxiangyu <huangxiangyu5@huawei.com>
|
|
|
|
JIRA: COMPASS-545
1. LVMISCSIDriver has been removed but LVMVolumeDriver is what actually exists.
Updating the LVM driver name in cinder.conf will solve this.
2. Add tgt to cinder-volume's service list
3. Add section of ceph in cinder.conf for cinder backend
4. Add enable_backends in cinder.conf
Change-Id: I477d2b17ce9e10871a7b96c1a98a49506b36a6ef
Signed-off-by: Yifei Xue <xueyifei@huawei.com>
|
|
JIRA: COMPASS-533
Bug:
When a vlan network is created and bind to a physical network,
the instances launched on this network will fail to get ip
address from openstack dhcp server.
Solution:
Set enable_isolated_metadata to True in dhcp_agent.in
Change-Id: I97cf8684d1ec9beeaa30e0ffa290150046ebcab1
Signed-off-by: huangxiangyu <huangxiangyu5@huawei.com>
|
|
JIRA: -
Sometimes memcached does not start after installation finished,
which will cause issues like openstack dashboard open instance
console failure.
This patch add memcached restart notify during package installation
and add memcached to cron service-status-check task's config file:
/opt/service
Change-Id: I625899348fb447875fcca7beedb619b5490ce567
Signed-off-by: QiLiang <liangqi1@huawei.com>
|
|
|
|
|
|
|
|
JIRA: -
Openvswitch service in Ubuntu named openvswitch-switch, but it named
openvswitch in CentOS.
Change-Id: I2a4d3d0ae0a646971696bae798b623da059a3781
Signed-off-by: liyuenan <liyuenan@huawei.com>
|
|
|
|
JIRA: -
e.g.:
root@host1:~# ip link show eth4
6: eth4: <BROADCAST,MULTICAST> mtu 1500 qdisc noop state DOWN mode DEFAULT group default qlen 1000
link/ether ec:38:8f:79:fa:5c brd ff:ff:ff:ff:ff:ff
root@host1:~# echo $(echo $@ | cut -d' ' -f17)
ec:38:8f:79:fa:5c
root@host4:~# ip link show eth4
6: eth4: <BROADCAST,MULTICAST> mtu 1500 qdisc noop portid 0100000000000000000000000000003000 state DOWN mode DEFAULT group default qlen 1000
link/ether e0:36:76:d2:34:73 brd ff:ff:ff:ff:ff:ff
root@host4:~# echo $(echo $@ | cut -d' ' -f17)
1000
root@host4:~# echo $(echo $@ | cut -d' ' -f19)
e0:36:76:d2:34:73
Change-Id: I916c19807da724e9ba471376f81f3c763b9340d5
Signed-off-by: QiLiang <liangqi1@huawei.com>
|
|
Change-Id: I6242b9fc09749ca83f446ef5456eaad72f248012
Signed-off-by: rssaravana <saravanakumar.somaskandan@huawei.com>
|
|
JIRA: -
You can add bond according to deploy/conf/network_cfg.yaml.
Change-Id: I70f2f03581cf763dbaf7a8a47bdbd46b66620fcb
Signed-off-by: liyuenan <liyuenan@huawei.com>
|
|
JIRA: -
Add libvirtd configure to support vm live migration.
Test cmd:
nova live-migration <vm-id> <dest-host-name>
Change-Id: Ic7f47f1e13a7cd84a2d1c67089fedf76662bf412
Signed-off-by: QiLiang <liangqi1@huawei.com>
|
|
JIRA: COMPASS-521
Use a hard value to set number of workers for Openstack API
service. This way only fix E9000 failed temporarily. Database need
to update in a late patch.
Change-Id: Ie03f6a758e1e458567054470962b5be4f6ee630e
Signed-off-by: liyuenan <liyuenan@huawei.com>
|
|
JIRA: COMPASS-528
Add tacker package making scripts into repo/pip/ directory.
Compass will install tacker using pip in this release.
Upgrade tacker role including tasks, templates and vars.
Modify the installation method from using source code to using whl package.
Change-Id: Iead323c7b46d8d1df084ed868d0e01bc1b0f7e55
Signed-off-by: Yifei Xue <xueyifei@huawei.com>
|
|
|
|
JIRA: -
use the first author's company for license holder
use the latest year for license year
add license for yml,c,cpp,shell,python,java.
don't add license for template files such as config and j2
patch for auto add license:
https://github.com/Justin-chi/Lab/blob/master/add_license.sh
delete the files with the Eclipse License
Change-Id: I4f17da4ebb36a074c0b6722a2a4154e6c235d4bb
Signed-off-by: chigang <chigang@huawei.com>
|
|
JIRA: COMPASS-512
1. Correct the script of unarchiving odl package for rerunnable
2. Delete the useless code in the script of installing networking-odl
3. Correct the variable "odl_pip" in vars/main.yml
Change-Id: I5488dd4caedcbda735de24fd594207d66310e828
Signed-off-by: Yifei Xue <xueyifei@huawei.com>
|
|
JIRA: COMPASS-512
1. Install requirements of networking-odl
2. Adjust order of removing neutron l3 agent
3. Modify file head
Change-Id: I27869821dc2f534500ec96b48df7a52ce27b58d8
Signed-off-by: Harry Huang <huangxiangyu5@huawei.com>
|
|
JIRA: COMPASS-521
Use a dynamic value to set number of workers for Openstack API
service when the number of cpu more than 4.
Change-Id: Ic6482cfad9b5eba8fad9e56ea1a33f65e8c71001
Signed-off-by: liyuenan <liyuenan@huawei.com>
|
|
|
|
JIRA: COMPASS-521
Nova compute use osapi_compute and metadata api. Both of their worker
processes should be limited, otherwise mysql would not deal with those
processes.
Change-Id: Ib00b4731369c1c677639c9437ab83887550db572
Signed-off-by: liyuenan <liyuenan@huawei.com>
|
|
JIRA: COMPASS-523
Heat needs grant heat admin role and create a menber role, otherwise
functest smoke test will failed when create stack by heat.
Change-Id: I925b473861bf82becf119bfd490c38732814c336
Signed-off-by: liyuenan <liyuenan@huawei.com>
|
|
|
|
JIRA: COMPASS-512
1. Use package distribution-karaf-0.5.2-Boron.tar.gz
2. Add configuring ACL
3. Remove neutron-l3 agent in odl_cluster role
4. Fix ODL l2 metadata problem
Change-Id: Ib14cdf3539deeeb91ebfec0450a9e7f0228f7741
Signed-off-by: Harry Huang <huangxiangyu5@huawei.com>
|
|
JIRA: -
Default Horizon theme is ubuntu. If you want to use openstack theme,
you can run `export ENABLE_UBUNTU_THEME=False` before deploy.
And if you want to change theme after you installed openstack, you need
to run this commands:
1. access compass VM, password is root
`ssh root@192.168.200.2`
2. change path:
`cd /var/ansible/run/openstack_newton-opnfv2`
3. Edit the `group_vars/all` file, change `enable_ubuntu_theme` var:
`enable_ubuntu_theme: False`
if `enable_ubuntu_theme: True`, the Horizon theme is ubuntu, otherwise
the Horizon theme is openstack.
4. run ansible playbook:
`ansible-playbook -i inventories/inventory.yml HA-ansible-multinodes.yml
-t horizon_theme`
Change-Id: I5465107228af32a482879c8a058899494f33e2d0
Signed-off-by: liyuenan <liyuenan@huawei.com>
|
|
JIRA: COMPASS-522
Bug:
Connection error happens when using aodh command.
Curl aodh-api also returns empty reply.
Solution:
Use apache to be aodh api server. Disable aodh-api
service. Delete aodh-api restart in ext-network role
to avoid aodh-api start again.
Change-Id: I04a4af38325bcbe4a0b86ec3d3beb7c9f3abd73b
Signed-off-by: Harry Huang <huangxiangyu5@huawei.com>
|
|
Modification:
1.Add congress support in deploy/compass_conf
2.Optimize congress role to be more independant.
Separate congress role into install, keystone,
database, start.
3.Support both ubuntu and centos
Change-Id: I1c9a458addd84133652fb7c0563bc287d161d129
Signed-off-by: Harry Huang <huangxiangyu5@huawei.com>
|
|
JIRA: COMPASS-520
After update ansible version to v3, keystone_user module only support v2 API.
So we use Shell Commands now, but it will failed with high probability.
Those Shell Commands should be instead by ansible modules to manage Identity users,
projects or some other work like crate networks.
Change-Id: I63d38b4a811a9c063ac4404da72787f594411b53
Signed-off-by: liyuenan <liyuenan@huawei.com>
|
|
|
|
Demand:
1.Compass4nfv master code only support one OPNFV version
now, no need to keep previous version's repo scripts
2.Currently java packages are in /var/www/package/onos
directory which is unreasonable
3.repo.conf has a lot of download URL which can turn into
pip requirements
Modification:
1.Remove previous version's directory in compass4nfv/repo
2.Only leave java and odl in compass4nfv/repo/features.
Other features can be added when they are ready
3.Seperate URLs in repo.conf into compass-core and openstack
pip requirements. Therefore pip.tar.gz only contains pip
packages for compass-core and pip-openstack.tar.gz only
contains pip packages for openstack
3.Move java packages into /var/www/package/java directory
4.Add make pip wheel package function to support congress
5.Use ODL package distribution-karaf-0.5.0-Boron.tar.gz
6.Use python package netwoking-odl-3.2.0
Change-Id: Ifa818e7630ac4435e3bdb7756b83f80f50c24a76
Signed-off-by: Harry Huang <huangxiangyu5@huawei.com>
|
|
JIRA: COMPASS-521
Number of separate API worker processes for service. The default is calculated by
the number of CPUs available for best performance. But some pods have many CPUs,
so that some service API worker processes is too many to access database.
For example, huawei-pod2 has 47 CPUs, nova-api has 98 processes, cinder-api has
50 processes and heat-api has 33 processes, when aodh want to initialize the database,
mysql can not deal with this processes.
Change-Id: I00513883ff55cc28cd676c3085615aaac36eb744
Signed-off-by: liyuenan <liyuenan@huawei.com>
|
|
Change-Id: I97413575e1cac87da4ac99bc45e1c55777f38243
Signed-off-by: meimei <meimei@huawei.com>
|
|
JIRA: COMPASS-518
The WSGIScriptAlias file should be "/var/www/cgi-bin/ceilometer/app".
This file is copied from
"/usr/lib/python2.7/site-packages/ceilometer/api/app.wsgi" in CentOS
or
"/usr/lib/python2.7/dist-packages/ceilometer/api/app.wsgi" in Ubuntu
Change-Id: Idcc37b0ed6c1d4c6e39d821a02a6579c57b6f035
Signed-off-by: liyuenan <liyuenan@huawei.com>
|
|
|
|
JIRA: COMPASS-518
Make Centos Newton ppa.
Update host Centos version to 7.3.
Support Centos Newton Deploy.
Support Ceph Deploy.
Change-Id: I6dd110cc10d425a9d590e1dcb5777ade5cce96aa
Signed-off-by: liyuenan <liyuenan@huawei.com>
|
|
JIRA: -
Remove some templates because it is duplicated.
Remove some repo files because master support newton only.
Change-Id: I6651191c6edb2ef6ab5da109922a889a8cd58a78
Signed-off-by: liyuenan <liyuenan@huawei.com>
|
|
JIRA: COMPASS-519
Demand:
Openstack Newton doesn't create flavors defaultly. But
functest needs to run with flavors exist
Solution:
Add an ansible role to create flavors after openstack
deployed
Change-Id: Ib7ac56b3d42d9ddfcbefffc44ac92fa81342368a
Signed-off-by: Harry Huang <huangxiangyu5@huawei.com>
|