This post has NOT been accepted by the mailing list yet.
We have a 3 node setup for ELK setup and it used to process around 1M to 2M docs per day. The cluster was having heap issues, so we added one more node to the cluster and triggered and restart of one of the nodes of the cluster to clear out the almost full heap of the node. After the reboot of the node, some of the shards were left in unassigned state. To fix this we issued a reroute command on those shards to assign them to the nodes. After all the shards got assigned the cluster state became green. But the problem is, in those indexes the documents which were present earlier are not there. But these indexes still occupy the same amount of space on the disk. So is there anyway to fix this situation, by making ES actually pick up the data from the disk.
We're using ES 1.4.2, heap size is 8,8,8,15 G for the 4 nodes respectively.