I am facing couple of issues when migrating data from 1.4 to 6.4.2. Made few changes to make the settings and mappings to make it compatible with 6.4.2 ( removal of mapping types and using text and keyword appropriately instead of string datatype)
I am running ES 1.4 on 5 Node cluster with 2GB allocated to Elasticsearch where as for 6.4 I am using a 3 Node cluster with 16GB allocated to elasticsearch out of 32GB.
There are close to 1500 indices with one shard and one replica for each index.
After migrating to 6.4 using the reindex API, I observe that
- The disk space occupied by all the indices for the 6.4 Cluster is 10GB ( combined all the nodes)
where as it occupies only 3GB for the 1.4.4 cluster.
- 1.4 Cluster works only with 2GB allocated to ES where as we require close to 16GB for 6.4
Basically the 6.4 cluster requires considerable memory for elasticsearch and the size of the indices has also grown up significantly. Is there a major change in higher versions that is causing this issue or am I missing out on something?