* NOVA_ZIPBALL_URL can be set for a XS/XCP install to point to a
non-standard zipball location.
* PIP_USE_MIRRORS (default True) can be set to False to stop pip
from attempting to use mirrors (useful for when using an offline
eggproxy).
Change-Id: I01a9944fe4c13c8cc8323a703f598bc498491b73
All heat users will need to run heat/tools/nova_create_flavors.sh as an admin
user. This change runs nova_create_flavors.sh if heat is enabled.
This saves the hassle of switching to an admin user and running this every time
devstack is started.
Flavors are changed in heat_init, so heat_init is deferred until after nova
is running.
Change-Id: I4576c83f5ef55809567e40e56c25eb0e1bbe6d45
This change lets the developer running devstack
control the token format used by keystone through
setting KEYSTONE_TOKEN_FORMAT in their localrc
file.
Change-Id: Ic1265fcb10b8de112891f61d5e07312322148ec2
Signed-off-by: Doug Hellmann <doug.hellmann@dreamhost.com>
Currently in devstack, Q_PLUGIN can be set to only one of well-known quantum plugins ("openvswitch", "linuxbridge", "ryu").
Setting it to a custom plugin causes stack.sh to exit with an error.
This commit lets you run devstack with Q_PLUGIN set to a custom Quantum plugin.
Change-Id: Iafe07cf162b5b07bee120b0c571149c374004bb3
Signed-off-by: Takaaki Suzuki <suzuki@midokura.com>
Fixes bug 1071567, the broken setup of external gateway IP/routes
when using quantum with the openvswitch plugin due to a misspelled
variable name.
Change-Id: I950701a55f770cf075bca500e5a935628f6bf14a
If ceilometer is enabled, nova instance usage auditing and the
appropriate notification drivers should be automatically enabled,
as opposed to relying on manual reconfiguration and service restart.
Change-Id: I41643a1418a87942be7b2b7979797ff5eb7e5479
fixes bug 1032022.
Devstack will not update service's start up command if the command already
exists in screen rc files due to previous deployment. There is no way to
clear a service's start up command so this would be a problem if the service's
start up command changes between different deployment. Clear the content of
screen rc file every time you run stack.sh to deploy openstack can solve
this problem.
Change-Id: I18d0700bb2169bdb0dadfa7982168aef2eff8b22
FLAGS.image_service was removed in:
[3aaa0b103447d56f8d3b259c693cd9a3a8dcbe36] Cleanup of image service code
Change-Id: Ie566f972e2c196228cd83edda0924b5a679a63a6
Puts additional state and lock paths for swift, quantum and nova
into $DATA_DIR/project that can be set independently of $DEST.
Change-Id: I2cca701856d3caa7fe472fefdd8b070a3c7f3adf
Raring is the new codename for the next Ubuntu release.
Change-Id: I53b5b4d23c0974427fbf0026a55f7b98b0fc3c76
Signed-off-by: Chuck Short <chuck.short@canonical.com>
It's used if installed and not used if not installed, which is confusing.
So trying to install it should make things a bit less confusing.
It's available under the same name in both Fedora and Ubuntu.
Change-Id: I6785dd5e831e9cf2da2eae5b116a9f5859c68c6d
The rpm list is more complete than the debian list, so complete the
debian one based on the rpm one.
Change-Id: I707a16d3d2646b4b3d0bc200fed62e5e6743c030
Now that a bootable volume can be created directly based on
image ID, we can dispense with the complexity around using
a builder instance to acheive the same effect.
Change-Id: Ied1f6863a4dd21685e2f135841b9e2c4d499675f
This patch allows using the Quantum Ryu plugin.
Ryu plugin lets Quantum link Open vSwitch and Ryu OpenFlow controller[1].
Ryu OpenFlow controller is not Openstack component, but I added some
processing that is related with Ryu to stack.sh for the convenience of
the person who intend to try the plugin.
Instructions for using Ryu plugin:
1. Enable services: "q-svc", "q-agt", "q-dhcp", "q-l3", "quantum", "ryu"
2. Set Q_PLUGIN to "ryu"
3. Set an internal network interface name to connect br-int on plural
hosts to RYU_INTERNAL_INTERFACE (optional)
Example localrc:
disable_service n-net
enable_service q-svc q-agt q-dhcp q-l3 quantum ryu
Q_PLUGIN=ryu
RYU_INTERNAL_INTERFACE=eth1
[1] http://osrg.github.com/ryu/
Change-Id: Ic1da132fa421f1c70c10a319ee3239831b0f956f
This is a REST API in the same style as other OpenStack APIs.
This also creates a new endpoint for the REST API which uses the
serivce type 'orchestration'. The old endpoint now has the service
type 'cloudformation'.
This matches the pattern where the native openstack API gets a
generic service type while the emulated EC2 API gets a specific
type (eg, object-store, s3).
There will be breakage for the time period where only one of this
change and https://review.openstack.org/#/c/14263/ are approved,
since keystone will have the incorrect service type for that period.
Change-Id: I6a0d51a63da8017d375b4c065c4c9079dfca8fe3
Change: I3bbcc5b9f8a4da2fcdb9f6f70913c2d6bc6e2b9b
added some timing messages around the volume operations in
exercises/volumes.sh. This was a good idea, but some
useless timing info was added to failed cases, this patch
pulls that back out. The only parameter used to detect
failure is a timeout, so outputing the time elapses in
these cases is useless.
Change-Id: I609a803a7293aa8a8e3cec186984de59bfe9b409
add auto cloning of the tempest tree if it's an enabled service
just reduces one step in getting tempest up in a new environment
Change-Id: Ia8a2feee96f26dffe96c87d572a31735d90cdabb
Related to https://review.openstack.org/#/c/13101/
Add tempest test variables, so tempest tests will be configured with
proper default values.
Change-Id: Iec13ec3492cbfa6dcce665a4e0723f1b941ae88a
