How-To: Determining Non-query Download of Bucket Segments

If you are seeing a high number of reads from a bucket, even though no user queries are being executed, you need to confirm which segments and dataspaces are triggering the download. Typically, non query segment downloads are triggered by replication within the LogScale cluster.

To determine whether particular queries are downloading segments from bucket storage for a reason other than performing a query you can execute the following query on the humio repository:

class=*BucketStorageDownload*
| reason!="segment-modified"
| groupBy(dataspace)

This will generate a count of the segments downloaded according to their dataspace name. For example:

You may need to check individual log lines to determine the reason for the download.