The l2.pp makes mistakes when finds neutron node if node has custom
nodename. Fix it.
Change-Id: I4cb984b760b056a5ab2583bcf6d44d4a8c385e8a
Closes-bug: #1590403
In some cases Ruby uses pointer instead of copying something for
optimization. But it allows to side effects. It was cause of that bug.
After cloning one variable by var.dup the problem was fixed.
Change-Id: Ide58c184f4cb9b655e11a13afb8c05bfc5149ab9
Closes-bug: #1625092
The network and uplink mapping now should be set in the new format:
a) Multiple strings each for one cluster in a textarea item;
b) Uplinks-mapping is optional
c) possible data is:
1. Cluster:VDS
2. Cluster:VDS:TU1;TU2;TU3
3. Cluster:VDS:TU1;TU2:FU3;FU4
where TU -- teaming uplink,
FU -- fallback uplink,
there is no limitation for amount of uplinks,
uplinks shoud be splitted by ';'
Change-Id: Ibf40675bb048a15a70157598931fae211da04edb
- Update driver
- Add support in templates and manifests. Agent will get CA from vcenter
computes hash.
Change-Id: Ic41d93b95aa9f163284492da60c64e27e1de5c92
Implements: blueprint custom-ca-bundle-verify-vcenter-cert
When an user doesn't set the 'Use the VMware DVS firewall driver'
checkbox, the parameter enable_security_group should be false.
Closes-bug: #1597240
Change-Id: I33cb22713e98bf6a8515403c9d0e5a7a6d9990af
Sometimes after removing a controller node from environment the
corosync resource for dvs agent becomes in error state where that
resource nowhere can be launched. Interesting that it never happens
for new resources.
That fix added deletion existed resource. Thus the problem reduces to
the previous.
Change-Id: I89c14af7f34406679a55751143f91fa12a33955b
Closes-bug: #1596487
Add few options to dvs-agent's init for restarting it after death of
the agent.
Closes-bug: #1584787
Change-Id: I0f86395c3f10c1807793d05709c0a68f2c1a3537
There is the task openstack-network-plugins-l2 that works unwell with
environments with custom hostnames. The root of problem is incorrect
detection of neutron nodes. Unfortuantely it happens in core task and
for fixing that task should be replaced.
Change-Id: I663960befa4cfa82348b231728722d774482fb4e
Closes-bug: #1590403
Compute-vmware nodes use obsolete version of prototcol for auth.
Set v3 instead v2.0
Change-Id: I20dd663101d7635ec69aa404f20b10b3e10cfa4d
Closes-bug: #1563374
Installation the VMware_dvs driver to /usr/bin and
/usr/lib/python2.7/dist-packages provides simplifying manifests,
removes some stuff and reduces amount of actions.
Also by prebuild hook the fuel plugin builder will build and put in the
plugin the latest verstion of driver from github repo.
Change-Id: I548f4a6d0ac4b4725befb27e812546333833a8d9
This function works incorrectly where several VMware clusters
are assigned to a controller node. Fix that mistake.
Change-Id: Ied22cd3927920299cd1ba870d42cf8486d07d281
By default driver considers wrong neutron-dvs-agent file.
This commit fix that in-place. More rigth way is fix that file
in the driver. I'm going to do it sometimes later.
Change-Id: If4225b198492b495509746fe4a635b1d28a1d25f
The new agent-based neutron ml2 driver makes deployment process more
granular and complicated. Add new task and new manifests for setup
and configure agents.
Change-Id: Ibf264640718e21dd02f43e0a755ace994fee5693
There is the aply-nova-patch resource which patch some files. Due to
unknow reason it works on controller and does not work on
compute-vmware when the path parameter is not specified. This commit
add path in both cases.
Change-Id: Ie34c068a203dc78d1eacd799102a9107c4048d68
Closes-bug: #1526691
There is the patch for nova which collects port numbers. This putch
shud be on the same node where nova-compute process works.
Add this patch.
Change-Id: I379a4fae00bb1b4b0e41d769b2747781d5abd423
There are some changes in this patch-set:
1. remove task which brokes deploy with compute-vmware
2. replace bringing full files vif.py and vm_util.py to aplly the patch
3. according changes for new network data scheme
4. new driver.
Change-Id: Ide36c894d5122d97a5bee591502b5c146c0b1fc8
Initial support of Fuel 8.0. Change metadata and other stuff.
Completed according changes for support new, granularted neutron tasks
Change-Id: Id8580ef4a00cbc1d185bcb3da60c7f54e3bb2c67
The name of parameter was changed from neutron_url_timeout in 6.1 to
url_timeout in Fuel 7.0. It requires some changes in the deployment
script.
Change-Id: I63a50bae1793d8b938c9dc525733064c95467d96
hiera('primary_controller') doesn't work for primary controller when it
has other role.
Check presents /etc/primary-controller.yaml works correctly.
Change-Id: Ie4e144d7d0456c89c0740d91edca8c2bdc560d3e
Closes-bug: 1504047
(cherry picked from commit bb388d179c)
This commit provides more precision version of value for parameter
notification_topics in neutron.conf file.
Сloses-Bug: #1484029
Change-Id: Iff3540d5ba29de83c28ea271e7163b2e03bea796
Network creation was described under if $primary_controller
condition. But when primary controller has additional role
(cinder-vmware for example) on stage of plugin deployment astute.yaml
linked to cinder-vmware.yaml where primary_cotroller == false for each
cases. For this reason net04 never created.
Deletion of this checking fix this bug.
Change-Id: I1fc914d3b09f2ef2683d2b82fcf318a69913a920
There was a bug 1481071 which was fixed on driver side. This patch-set
does corresponding changes on the plugin level.
Change-Id: I7268c3201664190a086d69615c66309ac020e620