I have a number of dangling indices across the cluster which all contain data that is no longer required. Is it safe to simply delete the particular directories under path.data that these dangling indices refer to, whilst the node is running? Or do I need to stop the node first?
The best thing to do is allow the dangling indices to be imported and then delete them through the API.
I would not recommend making any manual changes to the contents of the data path. Certainly not when the node is running, but even when it's down it's a risky business.
Probably. Either that, or rename the existing one by reindexing and then delete it. Once I've done that, the dangling indices ought to get imported, and can then be deleted, right?
Not sure. These indexes, and the log messages, originate from a time before I started working at the company.
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.