I have a elasticsearch cluster that we just migrated from 5.6.4 to 6.8.10. The cluster has 3 master nodes, 3 client nodes and 3 data nodes. They have the following resources allocated:
- master node : CPU 1000MHz, Memory 2000 MiB
- data node : CPU 2000MHz, Memory 3000 MiB
- client node: CPU 1000MHz, Memory 2500 MiB
I am just performing a reindex operation using the Upgrade assistant 7.0 but is taking way too long and it will be hard to do the same thing in production. For example an index that had ~20GB with ~ 80 mil documents took ~1h.30min and we have multiple indices like this.
How can we improve the speed of the reindex process?