thanos: Compact deduplication and retention is not working as expected
Thanos, Prometheus and Golang version used: Thanos 0.22.0 Golang go1.16.6
Object Storage Provider: S3 (Minio)
What happened:
We have two Prometheus servers with the exact same configuration to scrape the targets, both with the same cluster
and different replica
external label and a compactor with the below flags:
--compact.enable-vertical-compaction
--deduplication.replica-label="replica"
--deduplication.func=penalty
--retention.resolution-raw=20d
--retention.resolution-5m=90d
What you expected to happen:
See only one block of data for a time range and no raw chunk older than 20 days, but here is the status of the blocks:
This block is obviously older than 20 days but it is still present in the object store:
About this issue
- Original URL
- State: closed
- Created 3 years ago
- Comments: 23 (11 by maintainers)
Replica label should be without “”
Can you enable the debug log?