When you add the node, the cluster will start to rebalance shards into it and fill it up as it goes by shard count, not disk size, and assumes all nodes are the same size (less than ideal, IMO).
When THAT node fills up, it won't allocate more shards to it but it'll all run normally, and add data, etc. Unless it totally runs out of space, then the affected indexes will go read only and you may have to move some shards.
Plus given the new one is 20T vs. the existing 25T, likely you won't fill up anyway, depending how full you are already.
Anyway, just add it and should be fine but keep an eye on it. It will fill slowly as it only moves two shards at a time and moving 20TB of data might take a while. As it gets close to the lower watermark (I think 85%), watch its behavior; it'll stop adding new shards, but existing shards can grow, though given shards should be <50GB or so, you'd need a lot to grow a lot to eat up your final 15%/3TB.
Apache, Apache Lucene, Apache Hadoop, Hadoop, HDFS and the yellow elephant
logo are trademarks of the
Apache Software Foundation
in the United States and/or other countries.