As expected, all the shards moved off of the 'active' nodes and relocated to the nodes marked 'archive'
Now, the part that I am struggling with, is if I issue the command: curl -XPUT localhost:9200/alpha1/_settings -d '{"index.routing.allocation.include.node_type" : "archive,active" }'
I would expect the shards would be distributed across both node type active and archive. This is not the case. All the shards are moved back to strictly the 'active' nodes. I don't believe I have any other allocation filters defined in the cluster.
index shard prirep state docs store ip node
...
alpha1 5 p STARTED 0 130b 192.168.1.231 mualelsp10-01
alpha1 5 r STARTED 0 130b 192.168.1.230 mualelsp11-01
alpha1 3 p STARTED 0 130b 192.168.1.231 mualelsp10-01
alpha1 3 r STARTED 0 130b 192.168.1.230 mualelsp11-01
alpha1 6 r STARTED 0 130b 192.168.1.231 mualelsp10-01
alpha1 6 p STARTED 0 130b 192.168.1.230 mualelsp11-01
alpha1 1 p STARTED 0 130b 192.168.1.231 mualelsp10-01
alpha1 1 r STARTED 0 130b 192.168.1.230 mualelsp11-01
alpha1 4 r STARTED 0 130b 192.168.1.231 mualelsp10-01
alpha1 4 p STARTED 0 130b 192.168.1.230 mualelsp11-01
alpha1 2 r STARTED 0 130b 192.168.1.231 mualelsp10-01
alpha1 2 p STARTED 0 130b 192.168.1.230 mualelsp11-01
alpha1 0 r STARTED 0 130b 192.168.1.231 mualelsp10-01
alpha1 0 p STARTED 0 130b 192.168.1.230 mualelsp11-01
...
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.