9.1 release uses custom package provider (`apt_fuel`) which can install
non-signed package from plugin repository. Plugin uses standard package
provider (`apt`) which fails to install non-signed package, due to this
deployment process stops. By adding '/etc/puppet/modules' to parameters
of task we can use `apt_fuel` provider for package installation in
plugin manifests.
Change-Id: I7e2f9b8fcc2f080d76761a4a24dca3882e17dd78
In order to apply timeout for query to the neutron service need to
restart the service nova-api.
Change-Id: Ic0684a3e751352b62d821ad1db811d315b0d553e
Closes-Bug: #1596000
Metadata router need some time to start, if we start neutron-servers on
controllers right after primary-controller thay may fail (unrecoverable
failure). This patch moves neutron-server start to post-deployment stage
giving metadata router to be ready to serve requests.
Remove ruby script that tries to check metadata router status.
Closes-bug: #1587074
Change-Id: I09edb236ed1a71d7a90e5d556e55fe22a67f716c
Occasionally a situation may arise, when the neutron-server run
simultaneously on different nodes - to avoid this problem is added to
openstack-network-neutron-start task parameter "strategy: one_by_one"
Change-Id: Idea8bc0bcac5b6407cb6e14cc0c97ee76c3dd815
Closes-bug: #1587814
* Deleted post_deployment tasks
* Added reexecute_on for netron tasks
* Added stop neutron service to apply the new settings
Change-Id: I522b7a25f1c9fec154b6074d610ed81c4a0684d5
* shared variables moved in class nsxv :: params
* from the module nsxv removed classes that do not use templates or
external files(from directory "files" module)
* hiera calls from classes replaced to parameterized classes
* fix keystone auth url and neutron url for compute-vmware role
* more "cosmetic" fixes
Change-Id: I24da204d8f433439cc6f60501d479c84cd3c21c3
Metadata proxy can accsess to metadata server via:
1) Public network from routed ip(need set ip,netmask,gw)
2) Management network from routed ip(need set ip,netmask,gw)
3) Management network from management ip(automatic allocation)
* Add checkbox to choose route to metadata server via public or
management networks.
* Add checkbox to reservation ip address in management network
for use with NSXv metadata proxy.
Change-Id: I71675dd0871266bf6efcbbbbb2b3d2d49c4944f4
* Adapt tasks for work with "task base" deployment
* Changed [neutron]/timeout in nova.conf to 600s
* Changed in the method of get the cluster moref ID to search for the
name of the cluster in a subfolder
* Added 'reexecute_on: [deploy_changes]' to post_deployment task for
reconfiguration nsx.ini after new compute-vmware node added
* Moved post_deployment from task.yaml to deployment_tasks.yaml
* Disabled getting upsream puppet modules, replaced by the use of
/etc/puppet/modules
* Override "neutron_qos" option to false
Related-Bug: 1560377
Change-Id: Iaf7656660a93011b0766743b1562d461fa694d67
- Specify upstream modules that plugin use in Puppetfile; right now
this file is taken from fuel-library/ repository on 05 Feb 2016
- upstream modules that plugin use are:
puppet-neutron
puppet-nova
puppetlabs-stdlib
puppetlabs-inifile
puppet-openstacklib
puppet-keystone
- for modules update we utilize modified update_modules.sh script
from fuel-library/
- add remove_modules.sh script that will clean repository
- hook update_modules.sh and remove_modules.sh into pre_build_hook
- add note about puppet-librarian in plugin build guidelines
Change-Id: Ib38037c77a981d74c4da0e8d74652b51947e2203
* Delete task which reconfigure gw on node with compute-vmware/cinder-vmware
roles. This not need after merge patch from this bug
https://bugs.launchpad.net/fuel/+bug/1544067
Change-Id: I7519edb7f9b74510459fabdac47d5e31f251e99b
* Added get vcenter cluster name from "Vmware" tab
* Added convert cluster name to morefid
* Added public/private key to controller/compute nodes
* Added script, what check and configure neutron, if need
* Added task for connect to controller from compute-vmware via ssh and
execute neutron reconfigure script
* Update documentation
Change-Id: I0dbd6b15fc2cb1efd0de009ee79a27222fe4bab3
* Added "skip" for new primary-* tasks(tasks for primary-controller)
* Added restrictions for hide plugin with neutron and vlan segmentation selected
* Change script to configure neutron haproxy config
* Added refresh_on for openstack-network-neutron-start task
Change-Id: I0a487ec5a81706e853a6ddcceb696fe5f4eaf7f0
- Introduce new deployment task that will increase server and client
timeouts.
- Default timeout (1m) that is configured by haproxy globally for
backends is not enough when Neutron is used with NSX plugin. For some
actions like router and network creation it takes more that 1 minute
to finish operation. Default value leads to 504 (Gateway timeout) that
neutron command line client receives.
Change-Id: I03fde105a737058e312620e9eb33412902a047e7
Closes-bug: #1508971
* Plugin removes restriction on 'Neutron with tunnelling' in cluster
creation wizard
* Plugin ships two .deb packages:
python-vmware-nsx - Neutron NSXv plugin
python-nova - Nova package with NSXv support in nova-compute
https://review.openstack.org/#/c/209372/https://review.openstack.org/#/c/209374/
For build plugin need fpb version 3.
Change-Id: If6ba42abd7aaabca7a02c45718a82c6f8ca1e5a2
Andreas Jaeger