7bc4528e9d
Using the upstream implementation of cinder backup and restore breaks bringing up any instance that was booted from volume and still running at the time of the backup. This is due to cinder restore process, writing data through iscsi back to the volumes, thus adding new iscsi options for the connection and breaking the existing connections to VMs. Changes done to fix this: - at restore, restore the iscsi config file, but also make a copy of it - after volumes are restores, use 'config_controller --restore-complete' to overwrite the broken config file, with the original one from the above copy - 'config_controller --restore-complete' replaces 'config_controller --restore-compute' on AIO setups - 'config_controller --restore-complete' must now be also run on non-AIO setups. - as part of the restore procedure, compute hosts cannot be unlocked until 'config_controller --restore-complte' is run Story: 2002824 Task: 22741 Depends-On: https://review.openstack.org/584492 Change-Id: I441997019987d593872c82dd62baafbb397a97b1 Signed-off-by: Jack Ding <jack.ding@windriver.com> |
||
---|---|---|
.. | ||
centos | ||
controllerconfig | ||
.gitignore | ||
PKG-INFO |