Clarify that RFE bug reports should not have an importance set
Filling the Importance field [1] with anything other than Wishlist for RFE bug reports
is wrong. People confuse priority with Severity (which is what the importance field is
about), and that can cause us some grief.
Therefore let's state clearly that from now on, RFE bug reports will be marked 'Wishlist'
only.
rossella [Wed, 11 Nov 2015 19:08:36 +0000 (19:08 +0000)]
Remove TEMPEST_CONFIG_DIR in the api tox env
Setting TEMPEST_CONFIG_DIR in tox.ini is breaking the gate,
probably because of a tox bug [1].
Removing the line to unblock the gate. This implies that
for local execution of the tests TEMPEST_CONFIG_DIR
should be set manually.
Ryan Moats [Wed, 11 Nov 2015 18:51:21 +0000 (12:51 -0600)]
Lower l2pop "isn't bound to any segement" log to debug
Per the discussion in bug 1362242, the log statement doesn't
appear to be associated with anything bad happening. Therefore,
this patch lowers the log level from WARNING to DEBUG to avoid
polluting operator's logs.
Change-Id: I55b6d1d30ed2ed55659c9a6809531cfcbc57f17c Signed-off-by: Ryan Moats <rmoats@us.ibm.com>
Closes-Bug: 1362242
Jian Wen [Wed, 11 Nov 2015 03:32:20 +0000 (11:32 +0800)]
More graceful ovs-agent restart
When the neutron server is down/busy the agent is not able to get any
port info. After the agent restarts, it will not install any new flow.
Cleaning the existing flows will break all networking until the agent
succeeds to sync with the neturon server.
This patch ensures the agent cleans the stale flows only after it
succeeds to sync with the neturon server.
Pavel Bondar [Wed, 28 Oct 2015 09:22:52 +0000 (12:22 +0300)]
Add call to pluggable IPAM from ml2 delete_subnet
ml2 plugin overrides delete_subnet and do not call super(),
so pluggable IPAM action defined in db_base_plugin_v2 are not called.
As a result subnets can not be deleted from IPAM driver if ml2 plugin
is used.
Added ipam.delete_subnet call into ml2 delete_subnet.
Patch includes UT to verify IPAM driver is called on subnet
allocation and deallocation.
Test class is inherited from TestMl2SubnetsV2, so all SubnetV2 tests are
additionally executed for ml2 plugin with reference IPAM driver.
Update internal snat port prefix for multiple IPv6 subnets
In neutron when multiple IPv6 subnets on the same
network are added to a router interface, the router
interface port is updated with multiple fixed_ips rather
than creating a new port. In the case of IPv4 subnets
a new port is created for every added interface.
For dvr routers there are internal snat ports that are
created for every router interface that is added to a
router. So when multiple IPv6 subnets on the same network
was added to a dvr router, the internal snat ports were not
appended with the IPv6 prefix properly.
An error log message was thrown for the SNAT port not
matching the given internal port for the IPv6 address.
This patch will address this issue when IPv6 subnets are
added or removed from the dvr router that are on the
same network.
Cedric Brandily [Sat, 7 Nov 2015 21:36:59 +0000 (22:36 +0100)]
Reduce duplicated code in test_linuxbridge_neutron_agent
This change defines and uses get_linuxbridge_manager method which
mocks some methods in order to instantiate and returns a
LinuxBridgeManager instance, this method allows to remove some
duplicated code from test_linuxbridge_neutron_agent.
Russell Bryant [Mon, 9 Nov 2015 14:31:36 +0000 (09:31 -0500)]
Document the neutron-release team.
Gerrit includes a neutron-release team. Document what this team means
today. Also propose that this team have merge rights to *all* Neutron
git repos so that a small group has the ability to quickly unblock
things if needed, which is especially important around release time.
Change-Id: Id3d4d051278930841f64a4153e40a2dd3d893e44 Signed-off-by: Russell Bryant <rbryant@redhat.com>
yan.haifeng [Mon, 9 Nov 2015 02:27:54 +0000 (10:27 +0800)]
Fix misuse of log marker functions in neutron
some log marker functions misuse in neutron,
ref http://docs.openstack.org/developer/oslo.i18n/guidelines.html
and http://docs.openstack.org/developer/oslo.i18n/usage.html
Henry Gessau [Thu, 5 Nov 2015 16:47:13 +0000 (11:47 -0500)]
More instructions for neutron-db-manage revision --autogenerate
More detailed instructions for --autogenerate, including how to rely
less on devstack. Also some tweaks to scripts and requirements to
simplify the process.
Akihiro Motoki [Sat, 7 Nov 2015 18:07:02 +0000 (03:07 +0900)]
Make command log in neutron utils.execute() a single line
command execution log from neutron.agent.linux.utils.execute()
consists of multiple lines, and the first line has less information
and it is hard to analyze logs.
Carl Baldwin [Mon, 5 Oct 2015 21:35:37 +0000 (21:35 +0000)]
Move update_fip_statuses to Router class
I noticed that ri.floating_ips was being accessed from the agent. It
didn't seem right. Also, noticed that this set gets passed around too
much between methods. It all boils down to saving the previous set
and then doing set difference to see which ones went away. This patch
set makes it simpler.
Martin Hickey [Wed, 4 Nov 2015 21:37:21 +0000 (21:37 +0000)]
Make '*' the default ml2 flat_networks configuration
Update the ml2 flat_networks configuration to set '*' as the
default because that has been effectively the default behavior due
to a bug with the empty list.
Also, fix the empty list configuration for disabling flat networks.
DocImpact
UpgradeImpact
Change-Id: I13eb22afd7c0d011154b58a1bd218fd4fa82a8d9
Closes-bug: #1437745 Co-Authored-By: Martin Hickey <martin.hickey@ie.ibm.com>
rossella [Thu, 5 Mar 2015 09:24:10 +0000 (09:24 +0000)]
OVS agent reacts to events instead of polling
OVSDB monitor generates the events that the OVS agent
needs to process (device added or updated). Instead of
polling the agent processes the queue of events.
This code only landed a few days ago. Updated the deprecated
message to be after Mitaka so that external plugins will have
some grace period to clean up their code.
Change function call order in ovs_neutron_agent during the
creation or modification of a port, in order to fulfill the
VLAN OVS tag information in the "port" register before calling
the SG agent. This information is needed in some SG agent
implementations.
Cedric Brandily [Tue, 3 Nov 2015 21:54:07 +0000 (22:54 +0100)]
Optimize delete_csnat_router_interface_ports db query
Currently delete_csnat_router_interface_ports provides an iterator as
filter to core_plugin.get_ports which is used in a SQL IN argument.
SQLAlchemy is able to avoid some db queries when the IN argument is
empty BUT not when the argument is an empty iterator and raises the
warning:
SAWarning: The IN-predicate on "ports.id" was invoked with an empty
sequence. This results in a contradiction, which nonetheless can be
expensive to evaluate. Consider alternative strategies for improved
performance.
This change replaces the iterator by a list in order to optimize db
queries when the filter is empty.
Kevin Benton [Tue, 3 Nov 2015 15:31:54 +0000 (07:31 -0800)]
Make string representation of DictModel generic
The previous version was specific to a port dict and would
fail if str() was called on other models. This iterates over
and logs all values of the dict instead so it works under any
dictionary.
Sean M. Collins [Mon, 2 Nov 2015 19:47:13 +0000 (14:47 -0500)]
Docs: clarify that AnySubnetRequest is optional
The phrasing of "may not support" was a little confusing, mostly because
I read it as an imperative - "you may not do this" instead of it
being "you can do this if you want"
Martin Hickey [Wed, 21 Oct 2015 14:24:13 +0000 (15:24 +0100)]
Update neutron-debug to use stevedore aliases
Updating neutron-debug as it is importing interface driver classes
directly and not using the load_driver_interface method.
Also adds unit tests for the neutron-debug tool.
AKamyshnikova [Thu, 8 Oct 2015 15:18:21 +0000 (18:18 +0300)]
Per-branch HEAD files for conflict management
We have pep8 check for validation revisions, but it allows
outdated changes go into merge queue. To prevent this added
CONTRACT_HEAD, EXPAND_HEAD files.