Age | Commit message (Collapse) | Author | Files | Lines |
|
VPP sometimes takes some time to come up, this causes VPP interface detection
to fail. This patch adds retries for interface discovery so we can properly
identify state of VPP interfaces.
Also contains minor fixes for regex for VPP configs generation code to prevent
cases where we might edit configs that are commented out.
Change-Id: I915d5455acb8d496438b9c9e851639d3a43e6fa9
Signed-off-by: Feng Pan <fpan@redhat.com>
|
|
Change-Id: I10939a9a9f93a0c3e5a140a0ef90e3c75118b6d1
Signed-off-by: Tim Rozet <trozet@redhat.com>
|
|
Until now, when we pass more than 2 DNS, only the first one is added in
the ifcg-* config files.
This patch-set proves the issue, and proves the correction actually works.
Change-Id: I70a779782ab87cd4f74f3c50b0e649f503b386e3
Closes-Bug: #1716171
(cherry picked from commit 8ddab87c6ee2e80025e9340d806f93921779198b)
|
|
The patch adds an OVS extra parameter to delete controller for bridges
configured with standalone fail mode. By default, bridges are created
without having an openflow controllers. If node is restarted, the bridge
is set to standalone mode but if a service managing the bridge sets a
controller, it will remain in the ovsdb.
As ovs-vswitchd sets the bridge behavior to normal MAC learning switch
only if bridge in standalone mode can't communicate with its controller,
leaving controller defined can cause node outage when bridge is used as
management network. In such case controller service, like
neutron-openvswitch-agent, would need to communicate over management
network but given that bridge is in standalone mode but communicates
with controller, management network won't be reachable. This creates a
chicken-egg problem.
By removing controller by default, ovs-vswitchd implements a normal
action rule to the standalone bridge and service can use the bridge as
management network and eventually set the brdige to secure and set the
flows manually.
See opened Bugzilla for more information:
https://bugzilla.redhat.com/show_bug.cgi?id=1473763
Closes-bug: #1712517
Change-Id: Iad48312667834ea8f5c7145595ae89cb5159b36d
(cherry picked from commit f8d76d2cdebfa0d06233a59a8f6539207c5b5a4e)
|
|
Change-Id: Ibcf9bac3e5863b624f533968bb5fa6fca466db94
|
|
|
|
The pci address of the DPDK NIC needs to be specified as dpdk-devargs
when adding the DPDK port. This is required for using DPDK with
OvS 2.7
Change-Id: I4975130961199ee04dd002ec987081299e3ddd4a
Closes-Bug: #1702457
Signed-off-by: Karthik S <ksundara@redhat.com>
|
|
|
|
|
|
So removing the documentation link.
Change-Id: I5167e5787206db429ac74c6c74d7c74da5f83375
|
|
The stop_dhclient_execution() method calls netconfig.execute
directly from the object, instead of from the instance of
the object. This has the effect that when execute is called,
noop is always set false, even if os-net-config was called
with --noop. This results in a failure or network outage.
This change checks the value of self.noop from within the
provider.apply() method before calling the
stop_dhclient_execution method.
Change-Id: Ie2cc4519d7028dceb3a14bd7711839834db0ebdf
Closes-bug: 1699590
|
|
This patch adds a jsonschema for os-net-config's configuration data
and a library function to validate configuration data based on this
schema. Adding schema-based validation allows catching a larger class of
errors (typos, missing required parameters, etc.) for all devices
configurable through os-net-config.
The validation is run in the os-net-config CLI after loading the config
file. If the config file fails to validate, the current default is to
just log a warning and try to continue. By providing the new CLI option
'--exit-on-validation-errors', this can be changed to log an error and
exist instead.
This validation is meant to be reusable, for example for pre-deployment
validation of network environments (see change
Ic16ee0bc353c46f8fe512454176a07ee95347346). Packaging with os-net-config
makes it easier to keep object model and schema in sync.
Change-Id: Ie4a905863b2d46c88d9cd6c3afc50e7d0a877090
Signed-off-by: Frank A. Zdarsky <fzdarsky@redhat.com>
|
|
'tox -e cover' fails as --coverage-package-name option for 'testr' is
not recognized. A work-around proposed by Infra team (see
https://bugs.launchpad.net/neutron/+bug/1454772) is to replace 'testr'
with 'test'.
This patch implements this change and adds the cover/ directory to
.gitignore.
Change-Id: I659445ea479ea80a4f21d6ebca896abb4fb104f2
Closes-Bug: #1703177
|
|
|
|
|
|
Os-net-config was not using the user supplied mapping file to map
interface names for interfaces in bonds or bridges. Fix is to set
the mapping file info when creating interface objects in bonds/bridges.
Closes-Bug: 1695122
Change-Id: I063256b93c44c2e7a05d9820b0c1f6bf5b73d7a0
|
|
|
|
This patch allows the Multiqueue setting for DPDK bonds. In case of DPDK
bonds, the Multiqueue setting needs to be done for each of the interfaces
attached to the bond.
Implements: blueprint ovs-2-6-features-dpdk
Signed-off-by: Karthik S <ksundara@redhat.com>
Change-Id: I21b46cee902a17f13df51d456648368e468aadb7
|
|
This patch allows the MTU setting for DPDK bonds. In case of DPDK
bonds, the MTU setting needs to be done for each of the interfaces
attached to the bond.
Change-Id: Ida627313d14a674430b2aff3644fd62b2e0bcab7
Implements: blueprint ovs-2-6-features-dpdk
Signed-off-by: Karthik S <ksundara@redhat.com>
|
|
Currently os-net-config exits immediately if an interface fails to
come up. This causes problems when we call it with a failsafe
configuration because it can result in working interfaces not
being configured if there are also non-functional interfaces, which
can leave a system unreachable over the network even though it may
have a working connection.
This change stores errors and handles them after all interfaces have
been processed to allow os-net-config to do what it can even with an
invalid configuration. If any failures are detected it will still
cause os-net-config to report failure at the end.
Change-Id: I3bc75e217d0b7c5ae62900f4253ad57ee3720685
Closes-Bug: 1692725
|
|
The multi-queue option for DPDK is applied via ovs-vsctl command
via ovs_extra params. This patch adds support for configuring the
rx_queue (multi-queue) value to the DPDK Ports.
Change-Id: Ib9faad5e9d49f78e3a0b45ef3ae0082f3e9d14a6
Co-Authored-By: Karthik S <ksundara@redhat.com>
implements: blueprint ovs-2-6-features-dpdk
|
|
This repo is not translated at all - and not setup for translation.
Remove Babel and Babel set up.
Change-Id: Ied2e0352f32108fed7b6e3e7355e9f4c029297f1
|
|
MTU value is already written to ifcfg file if provided on the interface,
but for dpdk port, we have to apply this to ovs-vsctl command via ovs_extra.
This patch is adding the support to configure the given mtu vlaue to the
dpdk port using ovs_extra.
implements: blueprint ovs-2-6-features-dpdk
Change-Id: Ic01ed8cee2edbd648de0b64dc7f01da80c153d81
|
|
|
|
|
|
vlan port is not supported on the netdev bridges. But the
dpdk bond sample contains this wrong configuration. Removing
the vlan port on the ovs_user_bridge.
Closes-Bug: #1674280
Change-Id: I1acc8a3b8bbacc613344207b43684b0f86ff4a15
|
|
Currently os-net_config will map nics from a user-supplied mapping file
only if the nic is active (operstate = up). This can cause problems
if a nic is in a bond and one of the bond's nics has no carrier.
This fix will map the nic from the mapping file if the nic is defined
on the system, regardless of the operstate status.
The fix implements a new function to return a list of available nics
(no check of operstate) for use if a mapping file is supplied. The
list of active nics must still be used in the default case when
numbering nics (no mapping file supplied). There is also some
cleanup to check if a user-supplied mac is in the mapping file before
attempting to convert the mac to a nic.
Change-Id: Ia5d8c8b49b7ac0b51ee42a754f06e5e53587a5f6
Closes-Bug: 1679787
|
|
|
|
|
|
|
|
By default coverage looks for os-net-config, but since that's not
a valid python package name the actual package is called
os_net_config. This change correctly specifies that.
Change-Id: I5949ad2380942440f74397c2d74c94bc6ef3c55d
|
|
It appears os-net-config is not in the sync list for g-r. We should
add it, but in the meantime this will get us up to date.
Change-Id: I828a394ce72824eaabe904b563f73734a88c21f8
|
|
Vector Packet Processing (VPP) is a high performance packet processing
stack that runs in user space in Linux. VPP is used as an alternative to
kernel networking stack for accelerated network data path. VPP uses DPDK
poll-mode drivers to bind system interfaces rather than kernel drivers.
VPP bound interfacees are not visible to kernel networking stack and
therefore require different configuration steps in os-net-config.
VPP interface will be used in Openstack by either Neutron ML2 driver
networking-vpp as an alternative to OVS, or by Opendaylight SDN
controller with Honeycomb agent.
This patch adds interface configuration support to os-net-config.
The kernel nic specified to be VPP interface type will be bound to
VPP with a DPDK poll-mode driver. Note that os-net-config will only
configure those settings that affect interface binding, all other
configurations for VPP will be configured throught TripleO Heat
Templates.
Implements: blueprint fdio-integration-tripleo
Change-Id: Iebb40b7c5b252c51e86b6f44bcf36ed206101390
Signed-off-by: Feng Pan <fpan@redhat.com>
|
|
'pass' is added on a valid function, which is not required
for a stub function which has some implementation.
Change-Id: I8c35790a8b2a378231df535663c3e6f3a5fe5491
|
|
Fixes an issue where string values for nm_controlled were not converted
to the appropriate boolean values.
Closes-Bug: #1673878
Change-Id: I912ca2ef54d1d1f97e76b3922b7c2fecc77040ff
|
|
|
|
This change adds a flag to the base interface type, nm_controlled.
If this flag is true, the ifcfg file will contain NM_CONTROLLED=yes.
Since this flag is applied at the base it applies to any interface
type.
Note that not all interface types are supported by NetworkManager
at this time, so this option should be used with caution.
A demonstration of the usage is included in the
linux_bond_networkmanager.yaml file in the sample directory.
Change-Id: I2df6ce5b4bdb04651f27fc5daa64aa752e47f3b1
Closes-Bug: 1671888
|
|
OvS needs to be restarted after adding a DPDK port. This change
shall be removed on migration to OvS 2.7 where DPDK Hotplug support
is available.
Co-Authored-By: Saravanan KR <skramaja@redhat.com>
Closes-Bug: #1668375
Signed-off-by: karthik s <ksundara@redhat.com>
Change-Id: I250bb8e9a8e50874a82bf7f159e5a9d5677ef5ac
|
|
|
|
|
|
This change allows the commands in ovs_extra to be expressed
as a string or as a list. If a string is passed, it will be
treated as a list with a single element, and other ovs_extra
elements will be appended to the same list, such as fail_mode.
In order for more than one command to be passed as a string,
the commands will need to be separated with double-dashes,
similar to the format used by OVS command-line utilities.
Change-Id: I3f69e7ab96ff6e06953c3838b3e187b93545d623
|
|
|
|
Network service is failing and not started when interface is binded
with dpdk driver and interface config file is available.
In such case, removing config file for the interface which is binded with
dpdk driver.
Change-Id: Id445fbfe7abdd54c2ea522960224c7f0e412dd30
Closes-Bug: #1657661
|
|
|
|
While creating DPDK bond, if-up of the bond interface
itself will activate the member interfaces. Individual
members should not be called for ifup.
Additionally, removing the option to select the primary
active-slave member for the bond, because of #1654975.
Once it is fixed, primary selection from the network
config will be added.
Change-Id: I83527ada4a3618927d55b7aa06d436686b12d5b8
Closes-Bug: #1643026.
|
|
This patch adds a check that ensures the ovs_extra option, if present,
is passed in as list and raises an InvalidConfigException if not.
It addresses the issue that a user may mistakingly pass the value as
string, which would cause an error later when appending the failure
mode or when formatting the ovs_extra parameter.
Note: Also fixes a sample file in which ovs_extra was passed as string.
Change-Id: I9e8e47390b63d284de10d27b1db2c2cc54c86924
Closes-Bug: #1654196
|
|
This patch improves Python 3 compatibility by replacing .iteritems()
with .items() for iterating over dicts and fixing two tests to use
portable string comparison.
Change-Id: I8e1acafe372f1696823561d6aa8aae5437d34025
Closes-Bug: #1654187
|
|
SR-IOV Virtual Functions will be present as an interface in the
directory '/sys/class/net'. As these are virtual interface created
during the deployment process, it has to be ignored in the nic
numbering logic.
Closes-Bug: #1653097
Change-Id: I118a7314d496b531c52be45521d393123cdfe915
|
|
|
|
H803 hacking have been removed.
https://github.com/openstack-dev/hacking/blob/master/setup.cfg
Closes-Bug: #1650741
Change-Id: I915f80e1ef5a04f627fae9d60644b191e9e1cbed
|