Why snapshot API requires shared file system?

Elastic Stack Elasticsearch
1

My case wanted to backup only one node of a cluster, I know it is good practice to backup the whole cluster, but my ELK infrastructure do not have a shared file system, if there is no other choice, of course I will create the shared file system, but I just wonder why I can't put my backup on local disk, what is difference between backing up in local disk and shared disk? Thanks.

2

If it's a single node then the local FS will be fine, as there are no other nodes to write from.

3

So I actually can snapshot 1 node in a cluster if I specify a path pointing a local FS?
Since I cannot find any prove on Elasticsearch official document or Google search.
Thanks.

4

Yep, only for a one node cluster though.

It's not documented as we recommend clusters being 3 or more nodes.

5

Thanks for answering.
How about 1 out of 3 node cluster? Is it possible to snapshot 1 node in the cluster if I specify a path pointing a local FS of 1 node?
Another question is what would be the "type" for the API request? (Official site only suggest "fs" for shared FS)

PUT /_snapshot/my_fs_backup
{
"type": "fs",
"settings": {
"location": "/mount/backups/my_fs_backup_location",
"compress": true
}
}

6

Nope, all nodes need to be able to write data to the storage location.

Just use fs.

7

thanks for answering,
one more further question
How much size would the backup be if my index is 10G in size? Or how large I should assume the size for the backup be?

8

It'd be a similar size.

9

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.