Sanity check please re: Splunk indexer remote storage configuration
(1) Our indexers are set up in a clustered environment, and originally frozen buckets were moved to a local external location using coldToFrozenPath.
(2) We are now trying to use a script to have all frozen buckets written to AWS, and the app containing the updated indexes.conf was distributed using the master node
(3) However, we are STILL getting "duplicate" buckets being written to the local storage instance, even though everything SHOULD only be written to AWS
(4) When trying to find out the 5W's, I've noticed that there is an /opt/splunk/etc/system/local/indexes.conf that ALSO exists on each indexer, and this version still uses the Frozen Path to the local storage.
Since the deployed peer-apps indexes.conf with the script does seem to be working correctly, is it safe (relative) for me to remove the /system/local/indexes.conf?
The peer-apps/xxx/local directory has higher precedence than etc/system/local on indexers.
To confirm the location of the setting that's overwriting your coldToFrozenPath, you can run btool on your indexers:
./splunk btool indexes list --debug | grep coldToFrozenPath
That should print out each line of all indexes.conf on the host that contribute to the final indexes config for the coldToFrozen option.
Find the line that has the wrong value and that's the file that you should update.
If you want to delete the system/local/indexes.conf file you should first check what it's doing.
To see what other config is being used from system/local/indexes.conf you can run:
./splunk btool indexes list --debug | grep system/local
Any config that is listed will be applied to that indexer from the system/local/indexes.conf file. Either move the config to a peer-apps app, or just remove it if you don't need it.