Add support for generically setting arguments for OpenStack Networking
for both the server and agent. This is done using Q_SRV_EXTRA_OPTS for the
server, which let you place settings in the "OVS" or "LINUX_BRIDGE"
sections of the plugin configuration file. For the agent, two variables
are defined: Q_AGENT_EXTRA_AGENT_OPTS lets you set items in the "AGENT"
section of the plugin configuration file, and Q_AGENT_EXTRA_SRV_OPTS
lets you set things in the "OVS" or "LINUX_BRIDGE" sections of the
plugin configuration file.
Implements blueprint devstack-quantum-config.
Change-Id: I9ba44b21231f88cda1bcc5e3389c7875f03d8145
Support is added for configuring devstack to use quantum's Modular L2
plugin, ml2. See https://review.openstack.org/#/c/20105/ and the
blueprint for details. Either the openvswitch or linuxbridge L2 agent
can be used with the ml2 plugin (or both in combination on different
nodes). Code configurating devstack for these L2 agents has been moved
into separate files so it can be shared between ml2 and the existing
monolithic plugins.
To use ml2 in devstack, set Q_PLUGIN=ml2 in localrc. By default it
will use the openvswitch L2 agent. To use ml2 with the linuxbridge
agent, also set Q_AGENT=linuxbridge in localrc. The local, flat, and
vlan network types are supported with either agent. Support for
openvswitch's gre network type is not yet implemented in ml2.
Note that ml2 does not yet return a useful value for binding:vif_type,
so nova's GenricVIFDriver cannot currently be used. If using the
linuxbridge agent, work around this by setting
NOVA_VIF_DRIVER=nova.virt.libvirt.vif.LinuxBridgeVIFDriver in localrc.
Implements: blueprint modular-l2
Change-Id: Ib3ed039d25295cf1ab268536cec408af8cd3795a
nova fake virt driver can be used to simulate running arbitrarily large
number of VMs. When the fake virt driver is selected, all other arbitrary
limits are disabled (quotas, scheduler limitations) as well. The fake
virt driver can be selected by adding the following line to your localrc
VIRT_DRIVER=fake
This setting can be used to do simulated large scale (as defined by
number of VMs) testing (such as boot 500 VMs at once) inside a single
devstack instance.
Change-Id: I000c3cba107cd91bef11c4d5325e1ce7308faa6c
ceilometer has dropped bin scripts and switched to console scripts
blueprint setuptools-console-scripts
Change-Id: If6dff10471e351fe3640a60cc45fef719f671997
Fedora (and RHEL) like to install things in /usr/bin, e.g.
01:30:42 Creating /usr/lib/python2.7/site-packages/nova.egg-link (link to .)
01:30:42 Adding nova 2013.2.a1031.gfc5137d to easy-install.pth file
01:30:42 Installing nova-dhcpbridge script to /usr/bin
...
The default nova/paths.py defaults to /usr/local/... which causes
quite a few errors when stressing various parts of nova and tools
aren't found.
Change-Id: Iaa93af96ddfb4deb6d16ea1dcac25ae6ed1e317d
This is essentially a revert of
73695d0ea490c4c7a1158957dd5a85586cfa0933. Since stackrc is
responsible for sourcing localrc, the required functions
are necessary (like for enable_service).
fixes bug 1186488
Change-Id: Iad90f802e77b94416821008c294c1a2ede8a4729
Set [database]connection option to mongodb rather than sqlite as mongodb
is the only fully supported db.
Change-Id: I2b9fbc65de4476593d456a782dbc7e6540a75141
Fixes: bug #1186240
The Nova v3 API will be disabled by default after 31164 goes through.
This change explicitly enables the v3 API in the nova config file
Change-Id: I8b4c64b4718fc96d7a928c6fa3a0e8716c7edba1
The kernel's cmdline was set during the initial devstack machine
installation. Thus, during second runs, the kernel's cmdline was not
updated. This patch extracts append_kernel_cmdline, and configures
domU's kernel cmdline every time. As some networking parameters are
passed through the kernel cmdline, this patch makes it possible to
change the network configuration, even if a cached devstack exists.
Related to blueprint xenapi-devstack-cleanup
Change-Id: I3b7175f4e83326c3e28825ac50625f6bd2a9a029
Simple check to see if EPEL repo is enabled for RHEL6 and, if not,
automated install of the repo RPM.
Also adds an additional sanity check which checking for pip. In the
prior case of EPEL being disabled, a RHEL6 host may not have access to
python-pip. Although this shouldn't happen, its a good sanity check.
Change-Id: I4e8a4dda9475b75d071534d4eef469198502a048
dnsmasq-utils is available on all Debian-based supported distros.
Debian Wheezy and Jessie need qemu to be installed in order for the resulting
OpenStack environment to be fully functional.
Change-Id: I910968f60ca1ed5c300a1fa599fb480d65591145
Per Bug https://bugs.launchpad.net/openstack-ci/+bug/1024487, this will
modify the rate-limiting setting for rsyslog.conf. If rate-limiting is
being used the addition will set it to 0 otherwise the addition will
add the variables to the configuration file and set them to 0.
Implements: Variables "$SystemLogRateLimitBurst 0", and
"$SystemLogRateLimitInterval 0" in "/etc/rsyslog.conf". These changes
are persuent to what has been outlined in the rsyslog docs as found
here : http://www.rsyslog.com/tag/SystemLogRateLimitInterval/
Updated commit implements changes in code and placement post code
review. Recent change implements the "SystemLogRateLimitBurst" in first
if statement, which was "SystemLogRateLimitInterval" and wrong.
Fixes: Bug 1024487
Change-Id: I2637889cbe9e5e87bbfc0f1bb5047abae34d953c
nodejs is required for Horizon (ships a copy of lessc internally). I
couldn't find nodejs in the rpm list, despite it being installed in
lib/horizon. This adds to the rpm list as NOPRIME and retains the old
install point.
RHEL6 has nodejs in EPEL now, so we remove the old hack that installed
an upstream version and add it to the install mentioned previously
Change-Id: Ib3ee2f7d77e22666c9055c8528288c01b46a95d3
Tweak devstack support to the nicira plugin in order to
reflect changes made in:
https://review.openstack.org/#/c/26427
This complements fixes for bug #1121605
Change-Id: I221466fb40412ceb457d3e888767291a76176aa6
rejoin-stack.sh would not specify screen session name which will lead
unstack.sh unable to kill the screen session started by rejoin-stack.sh
after hypervisor rebooted. So specify the session name if
rejoin-stack.sh need to start a new sreen session.
fixes bug 1182806
Change-Id: I68635887aa82fa63f3667e0f090ea2836eec7b68
Currently lib/heat says:
To enable, add the following to localrc
ENABLED_SERVICES+=,heat,h-api-cfn,h-api-cw,h-eng
Once the stack is up, `heat list` fails because the API server
isn't running.
This commit adds h-api to that list.
Also make sure h-api is killed in stop_heat().
Change-Id: I2e818bb343680b3778f9277c23c766f784d28887
* It's not possible to prevent python-nose (which is not compatible
with Tempest) from being installed since so many os packages
dependent on it.
* This patch installs python-nose1.1 from EPEL and adds a
/usr/local/bin symlink for nosetests1.1 so that invocations for
Tempest will work as expected.
Change-Id: I756d1ca5eda127dcdd2efb047c8e14dd344f710d
This patch makes it easier to support the all-in-one installation of
XenServer with Quantum:
- separate agents for domU and dom0 openvswitches. The domU agent's name
is "q-domua", and the dom0 agent is called "q-agt"
- create a physnet1 physical network to connect those openvswitches
Note: This patch depends on the quantum patch:
https://review.openstack.org/15022
For a diagram, see:
http://goo.gl/BuAdg
Change-Id: Iff851438553528b9051edd380ecbb2c0415cbf78
Switching to the redis matchmaker.
The localhost matchmaker does not presently work.
Also, the localhost matchmaker could not work for
multi-host setups.
Change-Id: I81a26b9af78328e360a18b87371c619e194365d2