We are using elasticsearch 6.8.5 version in production with 3 Master and 8 data nodes. There is no path.repo property present in the production. I can't update elasticsearch.yml and restart nodes in production.
My Question is
Without path.repo property, Can I use snapshot and restore api ?
As I said, We have 3 Master and 8 data nodes and consider I'm using below request to register snapshot repo
question is which node will have the back up? all the 11 nodes (with their own data backed-up in the "/home/esbackup" location) or any one of the node (mostly which I login and send a backup request) will have the entire cluster back up (in "/home/esbackup" location)
I believe path.repo need to be set on all data and master nodes in the cluster. All nodes work together to create the snapshot so all need to have it minted and available.
You can't, if you want to use a filesystem repository you need to have path.repo in your elasticsearch.yml, you will need a full restart to add it, because it needs to be present in all nodes.
Also, it needs to be a shared file system, all the nodes need to have access to the filesystem, you can't use a local path that exists only in one node.
The other option is to use a cloud based snapshot on aws, gcp or azure, but this will also needs a full restart since you would need to install the repository plugin.
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.