Replies: 1 comment 2 replies
-
I vote for this option. Seems like cleanest to me, at the cost of having two different object-store clients for two releases. Note that Thanos objstore client currently doesn't support recursive listing used by our own clients, so that is something we may want to implement there first. |
Beta Was this translation helpful? Give feedback.
2 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
As discussed in past community meetings there are two Object Storage client implementations used by the Cortex project. The chunks package implementation and the Thanos objstore implementation.
Currently, we use the Chunks client for:
We use the blocks client for the TSDB blocks backend.
I want to take up the work of migrating the Alertmanager and Rules Storage backends to use the same client as the TSDB blocks backend. The goal of this is to:
I wanted to use this discussion to figure out what the best course of attack would be for tackling #2262. Since this will be a breaking change I'm wondering how we could keep compatibility for two versions before deprecating the old config. For the GCS client this is trivial since the configs are very similar. For S3 there are some features in the chunk client that do not exist in the TSDB objstore client. How would we go about deprecating those features without having a breaking config change for two versions?
Some ideas I have include:
Beta Was this translation helpful? Give feedback.
All reactions