Restore

This section describes the process to restore the Solr collections and Information Store database in SQL Server of an i2 Analyze deployment in a containerized environment.

The examples/pre-prod/walkthroughs/backup-restore/restore.sh script is a worked example that demonstrates how to script the process of restoring the Information Store database and Solr collections.

Understanding the restore process

To restore the data and indexes for a deployment of i2 Analyze, restore the pair of Solr collection and Information Store database backups.

To ensure that the data is restored correctly, restore the pair of backups in the following order:

1. Information Store database
2. Solr collections

If any data has changed in the Information Store after the Solr backup, the Solr index is updated to reflect the contents of the Information Store database when Liberty is started.

Preparing the environment

Before you can restore the backups, clean down the Solr, ZooKeeper & SQL Server environment if you are not using a clean environment.

See the Removing Solr, ZooKeeper, and SQL Server section of the walkthrough script.

Stop the liberty servers

Before the restore process can begin you must stop the liberty servers. In a this can be done by stopping the containers

The following command is an example of how to stop the liberty containers:

docker container stop liberty1 liberty2

Restoring the Information Store database

The process of restoring the Information Store database contains the following steps:

1. Running a SQL Server container with a new instance of SQL Server
2. Creating the Information Store database from the backup file in the new instance
3. Recreating the required logins in the new SQL Server instance, and the users in the restored database

Running a SQL Server container

Run a container for the new instance of SQL Server, this will have the backup directory available. For more information about running a SQL Server container, see SQL Server.

See the Running the SQL Server section of the walkthrough script.

Creating the database from the backup file

In a new instance of SQL Server, the only user is the sa user. Use the run_sql_server_command_as_sa client function to run the following SQL command as the sa user to create the ISTORE database from the backup file:

run_sql_server_command_as_sa bash -c "/opt/mssql-tools/bin/sqlcmd -N -b -C -S sqlserver.eia,1433 -U \"\${DB_USERNAME}\" -P \"\${DB_PASSWORD}\" \
    -Q \"RESTORE DATABASE ISTORE FROM DISK = '/backup/IStore.bak;'"

See the Restoring the ISTORE database section of the walkthrough script.

Recreating the logins and users

In SQL Server, a login is scoped to the Database Engine. To connect to a specific database, in this case the ISTORE database, a login must be mapped to a database user. Because the backup is completed for the ISTORE database only, the logins from the previous SQL Server instance cannot be restored. Additionally, the database users are restored but there are no logins mapped to them.

For more information about SQL Server logins, see SQL Server Login documentation

In this environment, create the required logins and users by dropping the users from the database and recreating the logins and the users by using the create_db_login_and_user client function. The logins and users are created in the same way as when original SQL Server instance was configured.

For more information about creating the required logins, users, and permissions, see Configuring SQL Server in the deployment documentation.

See the Dropping existing database users and Recreating database logins, users, and permissions sections of the walkthrough script.

Restoring the Solr collections

Restoring the solr indexes includes the following high-level steps:

• Deploy a new Solr cluster and ZooKeeper ensemble
• Restore the non-transient Solr collections
  • Monitor the restore process until completed
• Recreate transient Solr collections
• Restore system match rules

Deploy a new Solr cluster and ZooKeeper ensemble

To restore the Solr indexes, deploy a new Solr cluster & ZooKeeper ensemble. For more information about running a clean ZooKeeper & Solr environment, see

• Running Solr and ZooKeeper.
• Create Solr cluster.

See the Deploying new Solr and ZooKeeper section of the walkthrough script.

Restoring the non-transient Solr collections

The run_solr_client_command client function is used to run the RESTORE API request. The restore operation must be performed for each non-transient collection that was backed up. The following curl command is an example that restores the main_index collection:

run_solr_client_command bash -c "curl -u \"\${SOLR_ADMIN_DIGEST_USERNAME}:\${SOLR_ADMIN_DIGEST_PASSWORD}\" --cacert /tmp/i2acerts/CA.cer https://solr1:8983/solr/admin/collections?action=RESTORE&async=main_index_backup&name=main_index_backup&collection=main_index&location=/backup/1"
"

To perform a restore operation, use the Solr Collections API. For more information about the Restore API command, see RESTORE: Restore Collection.

Note: The restore API request must be an asynchronous call otherwise the restore procedure will timeout. This is done by adding async flag with a corresponding id to the curl command. In the above example, this is &async=main_index_backup.

See the Restoring non-transient Solr collections section of the walkthrough script.

Determining completion of Solr restore procedure

The Monitoring Solr restore status section of the walkthrough script runs a loop around the get_async_request_status client function that reports the status of the Asynchronous request. For more information about the client function, see get_async_request_status.

Recreate transient Solr collections

Recreate the transient daod_index, vq_index, highlightquery_index, and recordshare_index Solr collections.

For more information about the creating Solr collections, see Configuring Solr and ZooKeeper.

See the Recreating transient Solr collections section of the walkthrough script.

Restore system match rules

After the indexes are restored, upload the system match rules file. Use the Solr ZooKeeper Command Line Interface (zkcli) to create the directory in ZooKeeper for the system match rules file and to upload it.

The run_solr_client_command client function is used to run the zkcli request.

The following command creates the directory in ZooKeeper where the system match rules file must be stored:

run_solr_client_command "/opt/solr/server/scripts/cloud-scripts/zkcli.sh" -zkhost "${ZK_HOST}" -cmd makepath /configs/match_index1/match_index1/app

Note: The path to where the system match rules file must be located is in the following format configs/<index name>/<index name>/app

The following command uploads the system match rules file to the directory in ZooKeeper created earlier:

run_solr_client_command "/opt/solr/server/scripts/cloud-scripts/zkcli.sh" -zkhost "${ZK_HOST}" -cmd putfile /configs/match_index1/match_index1/app/match-rules.xml /backup/1/system-match-rules.xml

For more information about the ZooKeeper command line utilities, see Using Solr’s ZooKeeper CLI.

See the Restoring system match rules section of the walkthrough script.

Start the Liberty containers

After the Solr collections and Information Store database are restored, start the Liberty containers.

The following command is an example of how to start the Liberty containers:

docker container start liberty1 liberty2

After the Liberty containers have started, the wait_for_i2_analyze_service_to_be_live common function ensures that the i2 Analyze service is running.

See the Start Liberty containers section of the walkthrough script.