Only global administrators and promotion administrators can change the deployment configuration. See more about privileges in the Authorization section.
The “Deployments” section of an environment dialog, allows you to configure how a deployment will work when this environment participates as the deployment target.
Before enabling deployments on an environment, you should have a good understanding on how they work. In the section Deploying Revisions there is a detailed explanation on deployments and examples of the different strategies working.
The common options that you can configure for any environment are:
- Enable deployments: Only environments with this option enabled admit working as the target of a deployment. The deployment will fail otherwise.
- Save backup in VCS when the deployment finishes: If selected, a backup of the deployment will be saved in the Version Control System you have configured. The backup will be performed both for successful and unsuccessful deployments, using the metadata of the first enabled server in the first enabled cluster.
- Execute rollback when the deployment fails in the first server: The first enabled server of the first enabled cluster of the environment works as a test base for the deployment. If this option is selected and the deployment fails there, the Solution Manager will restore its previous state by executing the backup saved during the latest successful deployment, in the Version Control System.
- Deployment type: There are two types of deployments according to whether
the Solution Manager interacts with the load balancer or not.
- In the Simple deployment mode, the Solution Manager only updates the metadata of the servers, one by one, without any interaction with the load balancer. This option is intended for offline applications.
- In the Without service interruption deployment mode, the Solution Manager disables Virtual DataPort servers or clusters in the load balancer and enables them again when the deployment finishes. To deal with the load balancer, the Solution Manager uses the scripts defined in the Deployment Scripts section.
When the deployment is configured as Without service interruption, you can choose among different strategies to perform the promotion. The actual options available depend on the topology of your infrastructure, which is shown by the parameter Number of clusters installed. This parameter shows the number of enabled clusters that have at least one enabled Virtual DataPort server in the environment.
In those environments with only one enabled cluster, the subsection Promotion policy when the environment contains only one cluster will be available.
In a deployment, a cluster is considered enabled if it is not disabled and it has at least one enabled Virtual DataPort server. In addition, a server is considered enabled if it is enabled and the cluster it belongs to is also enabled.
In such scenarios, you can configure the following options:
- Server promotion strategy: Solution Manager uses the scripts Enable
server in the load balancer and Disable server in the load balancer to
enable and disable Virtual DataPort servers in the load balancer respectively
(disabled Virtual DataPort servers are ignored in this process).
It can promote servers in two different ways:
- The Half strategy disables half of the Virtual DataPort servers in the load balancer first, performs the deployment and enables them again. Then, it repeats the process with the other half. This option ensures that the potential system inconsistencies will only last a short time.
- The One By One strategy disables the first Virtual DataPort server of the cluster (according to the order in the catalog tree of the Solution Manager) in the load balancer, performs the deployment and enables it again. This process is repeated server by server, so the system may present inconsistencies for a longer time than for the Half strategy.
- Secondary shared cache available: To avoid any problem with the cache of Virtual DataPort servers, Solution Manager allows you to define two different cache data sources between which to alternate on every deployment. As servers update their metadata during the deployment, their caches swap from the current (old) data source to the other one, which contains updated data and, maybe, views with updated schemas. In this way, those servers that have not been updated yet are not affected by the changes in the cache. For this to work, there have to be two different schemas in the same database that can operate as cache for Virtual DataPort.
- First cache URL: URL of one of the schemas that can work as the cache for the Virtual DataPort servers.
- Second cache URL: URL of the other schema that can work as the cache for the Virtual DataPort servers.
Currently, the cache swap process is limited to global cache, this is, there are no caches defined for particular databases.
In those environments composed by several enabled clusters, the subsection Promotion policy when the environment contains more than one cluster will be available.
This subsection only contains the option Cluster promotion strategy, which configures how clusters are disabled and enabled in the load balancer during the promotion. Take into account that the Solution Manager uses the scripts Enable cluster in the load balancer and Disable cluster in the load balancer, respectively, for this purpose (disabled clusters are ignored in this process). There are two strategies available:
- The Half strategy disables half of the clusters in the load balancer first, performs the deployment for the servers defined in those clusters and enables them again. Then, it repeats the process with the other half. This option ensures that the potential system inconsistencies will only last a short time.
- The One By One strategy disables the first cluster of the environment (according to the order in the catalog tree of the Solution Manager) in the load balancer, performs the deployment for the servers it contains and enables it again. This process is repeated cluster by cluster, so the system may present inconsistencies for a long time.
Data Catalog Server Synchronization¶
It is also possible to configure Solution Manager to synchronize Data Catalog servers.
The user configured for each Data Catalog server must have at least the data_catalog_admin role in order to be able to synchronize the catalog metadata during the deployment process.
The options that you can configure for Data Catalog Synchronization are:
Enable Data Catalog synchronization: Data Catalog servers will be synchronized only if this option is enabled.
Select the Data Catalog elements to synchronize: Select the type of elements to synchronize (same options as in the Data Catalog elements synchronization).
- Views: Synchronize views metadata.
- Web services: Synchronize web services metadata.
- All: Synchronize metadata of all previous types.
Data Catalog synchronize priority: After synchronizing, the schemas will be updated to reflect the current ones of the server, and the descriptions can be updated in three different ways:
- Server: the descriptions will be overwritten with the server ones.
- Local: all the descriptions of the Data Catalog will be preserved.
- Server with local changes: the descriptions will be overwritten with the server ones, except the ones that have been edited from the Data Catalog (that will be preserved).
Select the Data Catalog clusters to synchronize: Select the cluster(s) with Data Catalog servers to synchronize. Only clusters with at least one enabled Data Catalog server will be synchronized.
Logical server name to synchronize: Optional. You can specify a logical name of a server to synchronize (this corresponds to the name of a Virtual DataPort server in Data Catalog). You only need to configure this if in the Data Catalog server there is more than one Virtual DataPort server.
If logical server name to synchronize field is empty, the metadata for all the Virtual DataPort servers defined in the Data Catalog server will be synchronized.