Hello folks,
Last week we successfully migrated our Cassandra database to separate Cassandra + Elasticsearch clusters with 1 node each, thanks to the great support from Skyline.
We would now like to understand how the backup mechanics work and in the event of a system failure, how to restore the databases.
After reading the documentation, the following is my understanding on how the backups and restore work so I may be wrong and thus would like some clarification.
The backups for Cassandra are still done as usual (as a complete DMA backup package if “Full backup” is selected as backup type in system center) and stored in the path specified here:
However, on the Cassandra server itself, old backups (also called snapshots) are not removed automatically and need manual cleaning.
As for Elastic, unlike Cassandra, the data is not integrated in the complete backup package but has to be treated separately using the “Standalone Elastic Backup Tool”.
The following setting in "system center -> backup" only works if Elastic is installed on the same machine as the DMA and NOT in our case:
Restoring our Dataminer system works by using the restore utility in the taskbar. This does not restore the Elastic data, which must be done separately by again using the tool.
Is this correct?
Best Regards,
Max
Hello Max,
I believe this answers your question: https://community.dataminer.services/question/dataminer-backup-full-backup-with-remote-cassandra-and-elastic-db/answer/61662/
In short, the answer suggests to use the following tools to back up and restore external database nodes:
Standalone Elastic Backup tool | DataMiner Docs
Standalone Cassandra Backup tool | DataMiner Docs
According to the documentation, the Cassandra Backup tool only creates a snapshot on the database node. To prevent loss of data in case of server failure, the snapshot has to be moved to an external storage manually or via some custom automation on the database node. (for example, a shell script triggered by cron in Linux).
To restore the backup, the contents of the snapshot directory must be copied to the respective database table directory.
Hi Max, there are a couple of things I omitted for the sake of brevity, but now I see it’s important to mention them in this answer.
1. The standard DataMiner backup utility doesn’t include a backup of a remote Cassandra database. Therefore custom backup logic has to be provisioned (schedule, copying and cleaning up snapshots).
2. To restore a Cassandra snapshot, the tables which are restored must be truncated first (this is mentioned in the Docs page for the backup tool).
Hello Alexander,
Thank you, I was not 100% sure about Cassandra but this answers my question.
BR,
Max