Grafana Loki Upgrade from 2.6.1 to 2.8.2

Grafana Loki
1

Hi,

We are running on Loki 2.6.1 in our production env currently; can we upgrade from Loki 2.6.1 to 2.8.2 directly? or are there any compatibility issues? Any suggestions on upgrade approach please.

2

I think you can, but you’d still want to be prepared and test in advance.

Here is our standard run book for upgrading our Loki cluster:

  1. Read the upgrade guide before upgrading Loki.
  2. Compare configuration difference between the two versions, see if any configuration is deprecated or removed (you can find the command in the link above).
  3. Make configuration change as needed.
  4. Update Docker image version.
  5. Deploy.

And I’d highly recommend you to test the upgrade in a dev cluster first.

1 Like
3

Hi Team,
Recently we have migrated to loki-2.8.2 release with the tsdb_shipper storage configuration from boltdb_shipper storage config. The issue we have noticed is active_index_directory path where ingester pods would write index files in loki, the name of the folder we have given as /var/loki/tsdb-index, so it supposed get created with that name inside the ingester pod but it’s not. it got created with index as a folder name like " /var/loki/index."

Below config is part of tsdb_shipper storage.

Please help me if any other configuration needs to add, if not how this tsdb functionality works? what is the reason active_index_directory got created with different name.

note: just to let you know cache_location path has been created with the proper naming as /var/loki/tsdb-cache name only inside the ingester pods only index path is creating issue.

storageConfig:
    tsdb_shipper:
      active_index_directory: /var/loki/tsdb-index
      shared_store: s3
      cache_location: /var/loki/tsdb-cache
      cache_ttl: 168h