Due to related bug it may happend that HA router will become active on
one of the L3 agents after few minutes since it was created. And during
that time when it's in "standby" mode, VM can be spawned, it will try
to get metadata but haproxy in the router namespace is still not running
so there will be no metadata available.
That will end up with SSH authentication error as there will be no
proper SSH key configured in the VM.
Originally this patch was intended just as a workaround for the related
bug but I think it's worth to keep it even if that bug will be solved on
the Neutron's side as it may give us more clear information about the
real problem in the future if something similar will happen and HA
router will not be active on any of the L3 agents.
Related-Bug: #1923633
Change-Id: I8c8b7c11c63ffeee4f776695f32ae686793299b3
We need such a test in order to be able to catch issues like [1].
The test creates network and server, then launches a VM with a FIP.
Then connects to the FIP and pings a global ip address.
The test requires global_ip_address variable in tempest.conf.
By default the value is not set and the test will be skipped.
[1] https://bugzilla.redhat.com/show_bug.cgi?id=1834433
Change-Id: Ibb3db814ed9b42f28ced8378368a3d5c2e43d7b3
* switch from neutron.tests.tempest to neutron_tempest_plugin
* Cleaned up README.rst and setup.cfg
* Use neutron_tempest_plugin as a tempest plugin package
* Fixed gitreview
* Keeping flake8 Ignores in tox.ini as tempest plugin is
imported from neutron codebase.
Change-Id: I42d389836e72813fdeebc797a577f4a8ac2ee603
