Chunk Deletion Indication

Grafana Loki
#1

Is there any indication of how the chunks are deleted from S3 for the boltdb-shipper setup for Loki?

The storage use seems to be increasing without any indication of dropping, and we’re at 80M objects and increasing in the S3 bucket.

The Loki config has the following enabled, and the bucket is months old, so I can’t really explain it away as a config issue.

table_manager:
      retention_deletes_enabled: true
      retention_period: 720h
#2

Retention in Grafana Loki through the table manager is done via the object store TTL feature. You can also setup retention through the Compactor which works only with the boltdb-shipper store. This requires more config for the Compactor. See Retention | Grafana Labs which contains an example of the retention configuration for the Compactor.