Datastore¶
Introduction¶
Admin user needs to create datastore and its versions as required.
A datastore is typically created as a type of database, e.g. the cloud admin could create 2 datastores for MySQL and PostgreSQL, separately. For each datastore, there could be multiple datastore versions. For example, for MySQL database, Trove could support 5.7.29, 5.7.30 or 5.8, etc.
Note
Starting from Victoria, the datastore version number must be the same with the image tag of the specific database. To support MySQL 5.7.29, a new datastore version with version number 5.7.29 based on mysql docker image needs to be created.
A datastore version is always associated with a Glance image, either by image ID or image tags. If the image ID is not provided, the image can be retrieved by the image tags. The tags are used for filtering as a whole rather than separately. Using image tags is more flexible than ID especially when a new guest image is uploaded to Glance, Trove can pick up the latest image automatically for creating instances.
Create datastore version¶
When creating a datastore version, Trove will create the datastore first if it doesn’t exist. Different datastore versions can have the same name but different version numbers, or same version number but different names.
When using image tags, make sure the image with the tags exists before creating the datastore version.
Note
From Victoria release, all the datastores can be configured with a same Glance image but with different datastore name and version name.
To create a datastore version:
Create a trove guest image
Refer to Build images using trovestack
Register image with Image service
You need to register your guest image with the Image service as cloud admin. In this example, the image is assigned tags that will be used when creating datastore version.
openstack image create \ trove-guest-ubuntu-bionic \ --private \ --disk-format qcow2 --container-format bare \ --file $image_file \ --property hw_rng_model='virtio' \ --tag trove --tag mysql
Create the datastore version
openstack datastore version create 5.7.29 mysql mysql "" \ --image-tags trove,mysql \ --active --default \ --version-number 5.7.29
Load validation rules for configuration groups
Background. You can manage database configuration tasks by using configuration groups. Configuration groups let you set configuration parameters, in bulk, on one or more databases.
When you set up a configuration group using the openstack database configuration create command, this command compares the configuration values you are setting against a list of valid configuration values that are stored in the
validation-rules.json
file.Operating System
Location of
validation-rules.json
Notes
Ubuntu 18.04
/usr/lib/python3/dist-packages/trove/templates/DATASTORE_NAME
DATASTORE_NAME is the name of the datastore, e.g.
mysql
orpostgresql
.RHEL 7, CentOS 7, Fedora 20, and Fedora 21
/usr/lib/python3/site-packages/trove/templates/DATASTORE_NAME
DATASTORE_NAME is the name of the datastore, e.g.
mysql
orpostgresql
.Therefore, as part of creating a data store, you need to load the
validation-rules.json
file, using the trove-manage db_load_datastore_config_parameters command on trove controller node. This command takes the following arguments:Data store name
Data store version
Full path to the
validation-rules.json
file
This example loads the
validation-rules.json
file for a MySQL database on Ubuntu 18.04:$ trove-manage db_load_datastore_config_parameters mysql 5.7.29 /usr/lib/python3/dist-packages/trove/templates/mysql/validation-rules.json
Hide a datastore version¶
Sometimes, it’s needed to make a datastore version invisible to the cloud users, e.g when a datastore version is deprecated or creating a datastore version for testing purpose, to do that:
$ openstack datastore version set <version-id> --disable