I am running ES 1.7.1 on a 9 node ES cluster. I had bunch of indices with replica of 0. I restarted es process on one of the nodes. The shards that were on that node are not being detected anymore. _cat/shards returns "missing shards exception". I looked inside the data dir of that node, and can see all the shards on the physical disk, its just that the ES process is not able to detect it. Is there some lock or something left around by the old running process ? Or would a restart of ES process necessarily make the indices unusable when you have replica of 0 and restart one ES process ?
thanks
-Kunal
What do the ES logs show?
There was nothing in the ES logs.
I got around this issue, by increasing the replica count from 0 to 1 for one of the index and that made the newly restarted node recognize all the shards that were present on it. Very weird.
-Kunal