- Mais recentes
- Mais votos
- Mais comentários
Hi,
If you still keep the source data somewhere then you can re-index the data to the new cluster.
For snapshot, note that the Amazon OpenSearch Service takes hourly automated snapshots and stores them in a preconfigured Amazon S3 bucket at no additional charge. To use this automated snapshots for disaster recovery you could raising a support ticket.
On the other hand, manual snapshot gives you more control so you can migrate data to a different domain by yourself. You could trigger Lambda function to create manual snapshot on a regular basis, but you may need to raise a support ticket to disable the hourly automated snapshot to avoid potential conflict as you can’t take a snapshot if one is currently in progress.
Have you taken a look at this?
GitHub: aws-samples/aws-cross-region-elasticsearch-backup-restore
This project creates a State machine that does daily automated Backup Restore of OpenSearch indices from one domain to another in same or different region
Conteúdo relevante
- AWS OFICIALAtualizada há um ano
- AWS OFICIALAtualizada há 2 anos
Jiaheng Thanks for your reply. But setting cross cluster replication won't be more cleaner and also the possibility of losing data would be less? What is the best approach that is being recommended for disaster recovery for Elasticsearch?
Hi,
It will depend on your use cases and the architecture of the whole system. For example, if your applications can easily switch to a new OpenSearch cluster (assuming it is in a different region) and the latency is acceptable, then cross cluster is better as the downtime is less comparing to recovering from snapshot. In this case, you could also consider migrating the OpenSearch domain from N. California region to a region has 3 AZs so you don't need to worry about the one AZ down scenario.