Hot and warm buckets reside on our fastest disks, but when we move to cold it is shipped off to a slower disk. There is a command to manually force the hot to warm transition but is there one for warm to cold?
My real curiosity is what splunk actually uses to transfer the data to the cold disk? A certain protocol or port? Something like scp or is it a splunk built in utility similar to indexing?
As I understand it, Splunk invokes the "move" or "copy/delete" capability of the underlying OS.
There is no command to move warm to cold, but there are mulitple ways to control when buckets move:
- by setting the number of buckets in warm (maxWarmDBCount)
- by controlling the size of the db (hot/warm) directory (homePath.maxDataSizeMB)
You can find out more by looking at indexes.conf.spec on your local Splunk machine, or by looking at the same file on docs.splunk.com
Here is another documentation topic that can be helpful: Configure index storage