8. Backup and recovery

8.1. Data Lake

See HDL documentation on how to backup data from the lake.

8.1.1. Identifier sequences

When migrating data to a new installation (including new loaders and/or RabbitMQ instances), the identifier sequences should also be migrated: identifiers used in the lake are set by the loaders, and the loaders get the identifiers from the broker (see Identifier sequences). To avoid duplicate identifiers (resulting in errors and rejected messages) loaders need to be decommissioned (which will return unused sequences to the broker), and afterwards copy the sequences in the pond-seq queue to the new installation (replacing the sequences in the pond-seq queue from a new installation).