I have talked with @untergeek in https://github.com/elastic/curator/issues/881 because I thought that I had a problem with force merge and curator.
It seems that there is a problem in other part which is not curator, and looks like ES is doing something strange.
I'm trying to _forcemerge in our current cluster: 50 nodes, hot-warm architecture, big and small indexes, and big indexes with 42 primary shards and 1 replication.
We are testing it on small indexes first (30k documents, 30MB disk size, 5 shards), and _forcemerge does nothing. There is no output, and there is no process running behind. Nothing shown on the logs (neither on the host where the command is issued, nor in the master) It should not take more than a few seconds to merge segments for a 30MB index, right?
Can you guys help me about what can be going on?
curl -XPOST 'localhost:9200/wormhole_2017-02-03/_forcemerge?max_num_segments=1'