The FWaaS policy.json entries are being relocated into the FWaaS
repository.
Co-Authored-By: Nate Johnston <Nate_Johnston@cable.comcast.com>
Depends-On: Idd8993da78dbd779e2b3d38a4039dabf60d416d5
Change-Id: Iec36638a7a57cc14753e93c5d207853d4a3e3b6e
Change I22ff5a5a74527366da8f82982232d4e70e455570 changed
get_ha_sync_data_for_host such that if an agent requests a router's
details, then it is always returned, even when it doesn't have the key
'_ha_interface'. Further changes to this change tried to put this check
back in (Ie38baf061d678fc5d768195b25241efbad74e42f), but this patch
failed to do so for the case where no bindings were returned (possible
when the router has been concurrently deleted). This patch puts this
check back in.
Closes-Bug: #1607381
Change-Id: I047e53ea9b3e20a21051f29d0a44624e2a31c83c
The MTU of VLAN subinterfaces must not be greater than
MTU of the trunk port. This commit adds a check that
verifies that this is indeed the case.
This limitation comes from the restriction that Linux kernel imposes on
MTUs of untagged interfaces and tagged subinterfaces of these
interfaces. For details of the current state, see
https://bugzilla.kernel.org/show_bug.cgi?id=13467, namely comments 7
and 8. Once this limitation is overcome, it would be possible to remove
this check.
Partially-Implements: blueprint vlan-aware-vms
Change-Id: Ic11e9a2cde4336e8e1533c56b63ff7aaf2b7a510
Commit 60325f4ae9ec53734d792d111cbcf24270d57417 used the depracted
models_v2.HasId.
Change-Id: I69dcaf740a782a370923e0022607518358456503
Closes-bug: #1620220
This is the last of the standard attributes that needs
to be exposed.
Partially-implements: blueprint vlan-aware-vms
Change-Id: I15dcf7f4207fb22376967b57a4024a74b7ad0466
The DB model uses 'revision_number' so we should keep the API
consistent with that to avoid the hassle of writing translation
logic as we transition to OVO.
This is an API change, but the revision plugin didn't exist
in the last cycle so we don't need to worry about backward
compatibility or an API impact.
Change-Id: I445974b0e0dabb762807c6f318b1b44f51b3fe15
Uncaught exceptions in the core or L3 plugin layers can cause the
auto_allocate plugin to fail unexpectedly. This patch ensures that
any unexpected error is properly handled by cleaning up half
provisioned deployment.
Related-bug: #1612798
Closes-bug: #1619497
Change-Id: I3eb9efd33363045f7b2ccd97fe4a48891f48b161
This patch improves how the trunk plugin supports future
segmentation types and how it can handle multi-driver
deployments where each driver may have its own set of
supported segmentation types. It does so by:
a) removing the existing segmentation type registration mechanism;
this has become superseded by how drivers advertise their
capabilities and register themselves with the plugin.
b) introducing a new compatibility check that makes the plugin
fails to load in case of conflicts: for multi-drivers environments
a common set of segmentation types must be found for the plugin
to load successfully. In case of single driver deployments,
a driver can bring its own segmentation type.
Partially-implement: blueprint vlan-aware-vms
Change-Id: I0bd8b479c07945c65e14d4b59dfbcc1bc9a85a88
This patch ensures that trunk status is set to DOWN when its
parent port is unbound. This typically occurs when the instance
using the parent port is deleted in Nova.
Change-Id: I2673d2dd2d1dfd8eaabab66c009311cf4c143734
Partially-implements: blueprint vlan-aware-vms
MySQL doesn't like foreign key columns to be modified,
this adjusts the script to add all constraints and make
modifications before setting up the foreign key relationship.
Change-Id: I494758120c8a87fe584c781b928f8b9d3bac5291
Closes-bug: 1619696
In case ensure_namespace fails for any reason, the ip_wrapper attribute
is not created but destroy(), that uses this attribute, is still called.
TrivialFix
Change-Id: I4135f6564a180b4bd275312866d3dc6e4b9b63ca
The patch proposes adding a new binding_index to the
RouterL3AgentBinding table, with an additional Unique Constraint that
enforces a single <router_id, binding_id> per router. This goes a long
way into fixing 2 issues:
1. When scheduling a non-HA router, we only use binding_index=1. This
means that only a single row containing that router_id can be
committed into the database. This in fact prevents over-scheduling of
non-HA routers. Note that for the HA router case, the binding_index
is simply copied from the L3HARouterAgentPortBinding (since they are
always created together they should always match).
2. This sets the ground-work for a refactor of the l3 scheduler - by
using this binding and db-based limitation, we can schedule a router
to agents using the RouterL3AgentBinding, while postponing the
creation of L3HARouterAgentPortBinding objects for the agents until
they ask for it (using sync_routers). This will be a major
improvement over todays "everything can create
L3HARouterAgentPortBinding" way of things).
Closes-Bug: #1535557
Change-Id: I3447ea5bcb7c57365c6f50efe12a1671e86588b3
Propose a new extension named "timestamp_ext" to add timestamp to
neutron ext resources like router/floatingip/security_group/security_group_rule.
APIImpact
DocImpact: Neutron ext resources now contain 'timestamp' fields like
'created_at' and 'updated_at'
Implements: blueprint add-neutron-extension-resource-timestamp
Change-Id: I78b00516e31ce83376d37f57299b2229b6fb8fcf
Generate a new context object request-id for each reference
to self.context. This allows easier tracking of requests
in logs.
This is the L3 agent equivalent fix of
I1d6dc28ba4752d3f9f1020851af2960859aae520.
Related-Bug: #1618231
Closes-Bug: #1619524
Change-Id: I4a49f05ce0e7467084a1c27a64a0d4cf60a5f8cb
This method was difficult to reason about. This refactoring preserves the
original behavior but reduces code duplication and gathers all of the error
reporting logic in to one section so that future patches can more easily
optimize it.
Unit tests from these packages cover this area well.
neutron.tests.unit.extensions.test_segment
neutron.tests.unit.extensions.test_subnet_service_types
Change-Id: I074f34f3d8bc4135664a94dee11ed01c952d2f3e
Partially-Implements: blueprint service-subnets
A change that merged recently for DHCPv6 lease release
did not have the correct format for the lease file in
the unit test, so we started seeing these today from
the functional tests:
AddrFormatError: failed to detect a valid IP address from '[2003::2]'
We have to strip the brackets from IPs read from the
lease file.
https://review.openstack.org/#/c/301747/ was the culprit.
Closes-bug: 1619463
Change-Id: I0964fe9d00f8f3922041b12c352449e943b8f251
This patch adds the name and description column to the
networksegments table.
Change-Id: I7edc224e25e604dfd7613b945a4ca16d9e385760
Partially-Implements: blueprint routed-networks
MySQL doesn't like foreign key columns to be modified,
this adjusts the script to add all constraints and make
modifications before setting up the foreign key relationship.
Change-Id: Ifdb9e57b27aae21cf0ec7c8fec25683b9b0892cf
When the DHCP agent requests a port from a server, it asks
for fixed IPs from specific subnets. However, if the server
is doing auto ipv6 addressing, it may give it an extra IP
it never asked for. The previous logic was not expecting this
so it would fail to correctly setup the DHCP port.
This patch just ignores any IPs on subnets the DHCP agent
isn't servicing.
Closes-Bug: #1618674
Change-Id: I1c26c85520258b751fa98eee326bc2acd2a16860
During the development of patch [1], it became apparent
that the DB was rounding the microseconds part of a datetime
object. This caused inconsistencies between what would be
stored in the DB and what was returned to the user via the
API because the API uses strftime.
For an example, see the INSERT statement on L1-L2 and then
the subsequent retrieval of the same record on L43 of [2].
The microseconds were rounded up even though the mysql docs
say they will be truncated.
In order to ensure consistency across SQL drivers, backends,
and our API, this patch adds a TruncatedDateTime object which
truncates off the microseconds before they go into the database.
1. I78b00516e31ce83376d37f57299b2229b6fb8fcf
2. http://paste.openstack.org/show/565116/
Closes-Bug: #1619299
Change-Id: I47d5ad5a5cdc0cb0d61f0642616cccc8f341ed62