Are those "_analyze" api parameters also allowed when creating the snapshot, or not..?
If not then these are no use unfortunately.
I have tested with this analyze property
And seems like the problem is those very tiny files (few kb or mb in size). When these are backed up, file handles change very rapidly, since these are so tiny. And, although Elasticsearch has already opened/closed like N amount of these files, our storage thinks many of these files are still opened.
For example, when I tested like this, I could clearly see these file handles being increased like 20-30.
curl -XPOST 'http://localhost:9200/_snapshot/arendus-dtelastic/_analyze?blob_count=100&max_blob_size=1mb&timeout=120s&pretty';
Adding &concurrency=1, does not change anything.
On the contrary, when I executed the command with these parameters, I could hardly see any difference regarding open file handle count:
?blob_count=10&max_blob_size=200mb
So moving larger chunks of data, is no issue.
Is there any hint, how to limit those tiny files being created..?
Regards
Raul