hi, anyone knows what will happen if I wrote to a disk on node A, then node A fails, then I move the disk to another node B (new member of the cluster or existing member)...
will I be able to recover this data?
the 3 node cluster has no replicas, only primary shards
So to test this, start a 3 node cluster, through some data at it, kill a node, start a fresh node with path.data pointing in other node data location, check logs?
Any other ideas?
Are there any docs about it? I wasn't able to find anything...
It is not a supported or tested method, which would explain why it is not documented. I would recommend instead running with a replica configured as this would be a lot easier to manage sand less risky. This would also help against corruption due to disk failures, which moving disk does not.
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.