As discovered in tempest, when you create and delete volumes
quickly, and have a small volume storage pool, you can get cinder
scheduler to think you are "out of space" when you are not.
This is because cinder scheduler updates free space on it's
periodic job, which defaults to 60 seconds. We need control over
that value for the devstack gate otherwise we regularly get
overruns if we run too many volume tests quickly.
Work around for bug 1180976
Expect that this will get removed later if cinder gets a more
efficient way to update the scheduler for freed resources.
Also, don't be completely stupid about setting defaults....
Change-Id: I20e52e66fcc94b224476cdd14c88bd6981b4e617
The GlusterFS volume driver can be used with Cinder by setting the
following in localrc:
CINDER_DRIVER=glusterfs
CINDER_GLUSTERFS_SHARES="127.0.0.1:/vol1;127.0.0.1:/vol2"
Shares are <host>:<volume> and separated by semicolons.
Change-Id: Iaa105233c9fce8d8fda0a9ea447e045b8d010db7
This creates /etc/tgt/stack.d as a symlink to the cinder volumes
directory and includes it in /etc/tgt/targets.conf in a similar way
to /etc/tgt/conf.d
targets.conf only gets appended to when the symlink is created.
Any old /etc/tgt/conf.d/stack.conf is deleted
Change-Id: Ifb6412f6860ceb28b724644fb9b618638f552102
Fixes: bug #1072121
Move setup_develop() call into the install_XXX() function as the
distinction has no real purpose. This eliminates the configure_XXXclient()
function for Nova and Swift.
Also create install_cinderclient() as we will need it to handle dependencies
in a followup patch.
Chasing down another #$^%&&^^^$%%#$ prettytable version pin
Change-Id: I61d761bc57a31caf2ed0ade326ab236eb2bb7358
for files that don't start with a #! or end in .sh, the added tags
are nice for emacs users to automatically switch to the right mode.
Change-Id: If4b93e106191bc744ccad8420cef20e751cdf902
* Clean up interactive configuration
* Complete moving initialization of service-specific varialbes into the
service lib/* files.
* Cosmetic cleanups
Change-Id: Iea14359bd224dd5533201d4c7cb1437d5382c4d1
* add CINDER_MULTI_LVM_BACKEND parameter
* if CINDER_MULTI_LVM_BACKEND = True, creates 2 LVM volumes
- Name : stack-volumes and stack-volumes2
- Size : 5GB for each of them
Change-Id: I6e03b96e6b2aed0f0c166472300139fba9eaa68b
cinder-backup service has been added to cinder. This fix adds the
cinder-backup service. To enable this service, c-bak must be added
to ENABLED_SERVICES in the users localrc. To actually perform a
backup the user must enable swift or another service.
Change-Id: If4ef8d78081be572987e7c38c76ff57e82b4f744
Always display 3 digits for milliseconds (appends 0s)
Based on I3f1461839258be0723e2d3616ec225a830d13029
Change-Id: I48b3f3781b4d34ed6a5fb9e4e78cee919afda6c1
get_python_exec_prefix returns the path to the direcotry where python
executables are installed, that is /usr/bin on Fedora and /usr/local/bin
everywhere else.
It is used to properly locate OpenStack executables.
Fixes: bug #1068386
Change-Id: I228498ebe2762568d00757d065e37377ee2c8fb3
This enables us to use Sheepdog as a Cinder backend storage by setting
the CINDER_DRIVER environment variable.
Change-Id: I70cfb7f89ac3260d277fa160c457d220255de065
This commit also changes the following:
- Fixes Nova QPID module path
- Fixes a bug Cinder ZeroMQ RPC points to nova module
- Adds ZeroMQ setting for Heat RPC
qpid_is_supported is moved from functions to lib/rpc_backend.
This work is based on the work by Isaku Yamahata <yamahata@valinux.co.jp>
in https://review.openstack.org/#/c/19074/.
Change-Id: I45e21b1fb85e539213f5243764132a37906d7455
On many systems the requiretty sudoers option is turned on by default.
With "requiretty" option the sudo ensures the user have real tty access.
Just several "su" variant has an option for skipping the new session creation step.
Only one session can posses a tty, so after a "su -c" the sudo will not
work.
We will use sudo instead of su, when we create the stack account.
This change adds new variable the STACK_USER for
service username.
Change-Id: I1b3fbd903686884e74a5a22d82c0c0890e1be03c
Removing the cinder volume group breaks devstack installations
that share that volume group with other logical volumes. It also was
leaking loopback devices.
Change-Id: Ice4470e06e08ce49a0e1f82af70abcc015c91c20
Cater for the pending more general cinder support
for configurable volume wiping method at:
https://review.openstack.org/#/c/12521
This change is done here first so as to not
trigger a CI lockup when the referenced patch above lands.
When that's in place, we can remove the older
secure_delete config adjustment in a subsequent patch.
Change-Id: I73fe2e0d1cf2815ab6025121584951cb5ff56fa3
cinder currently has issues that leave volumes around
after tempest tests. Make sure that cinder gets cleaned
up to a zero state on an unstack.sh so that we can reset
the environment.
Change-Id: I448340899bf0fae7d4d16fa26da17feafcef888f
This prevents old invalid tokens from working after a rerun of stack.sh
and potentially providing users and tenants that don't exist.
Fixes bug 1089700
Change-Id: Icfc22978e41e459d51b50bc7ad2e6d98d766e402
Fixes bug 1088801
devstack does not create signing_dir when keystone token format is UUID.
If the default value of signing_dir is read-only, OpenStack services
such as Quantum server failed to start due to permission denied.
On the keystone client cannot know which token_format is used in keystone
in advance, so signing_dir should be created regardless of the token format.
Change-Id: I1b0d25c1ac4d22d9fb2c5443d15b96fdaa5a4c81
Between is_fedora, is_ubuntu and is_suse, we can make the code a bit
simpler to read. We also use exit_distro_not_supported to identify
places where we need implementation details for new distros.
As "/sbin/service --skip-redirect" is Fedora-specific, guard this with a
is_fedora test too.
Change-Id: Ic77c0697ed9be0dbb5df8e73da93463e76025f0c
Fixes bug 1087329
As the driver was moved to a different location, devstack script needed
an update.
Change-Id: Iaa1db94a84b6e9cb99514ce886025600809e9f29
On one system multiple volume and volume manger could be installed and
needs dedicated tgt config entries.
cinder-volumes, stack-volumes, nova-volumes are the default volume group
names.
/etc/tgt/conf.d/ files should be named based on the volume-group name.
The vg name is uniq on one system.
In devstack case the stack.conf is usable.
Changes:
* Rename conf.d/cinder.conf to conf.d/stack.conf
* Handle conf.d similary on all distribution
Change-Id: I856cdf4a21a414d2940d8f9d8b0b0368b1fad887
fixes bug 1087272
Trying to configure XenAPINFS volume driver through localrc options
failed. This fix removes the extra check, as lib/cinder already exits on
error.
Change-Id: I874b7cee44861244cb7a340cc4094ef3f8b48a5a
This replaces all of the [[ "$os_PACKAGE" = "deb" ]] tests, except when
those tests are before straight calls to dpkg.
Change-Id: I8a3ebf1b1bc5a55d736f9258d5ba1d24dabf04ea
Related to bp xenapi-storage-manager-nfs
Add configuration options to devstack scripts, so it is easier to set up
a system with a XenAPINFS volume backend. It makes possible to test this
configuration with exercises. To enable the XenAPINFS driver, specify:
CINDER_DRIVER=XenAPINFS
CINDER_XENAPI_CONNECTION_URL=<url for a xenserver>
CINDER_XENAPI_CONNECTION_USERNAME=<username for the xenserver>
CINDER_XENAPI_CONNECTION_PASSWORD=<password for the xenserver>
CINDER_XENAPI_NFS_SERVER=<a box that has an nfs export>
CINDER_XENAPI_NFS_SERVERPATH=<exported catalog>
in your localrc
Change-Id: Ia214172aac377d273a03849c8cc2adcbf5b8f607
Note that this is the first part of the support. A second part involves
dealing with the package names.
Among the changes:
- add several functions to determine some distro-specific behavior (how
to call usermod, if some features are available on the distro, etc.)
- correctly detect openSUSE and SLE in GetOSVersion, and set DISTRO
accordingly
- new is_suse() function to check if running on a SUSE-based distro
- use zypper to install packages
- adapt apache virtual host configuration for openSUSE
- some simple fixes (path to pip, mysql service name)
Change-Id: Id2f7c9e18a1c4a7b7cea262ea7959d183e4b0cf0