Object storage (AWS S3, Azure Blob Storage) support in Helm charts

[asakth22]

I'm using Jaeger with Elasticsearch as backend with persistent volume.
But I would like to store the data backed by object storage.
I can't find options in the helm chart.
So any thoughts on this ?

[dosubot[bot]]

Hey @asakth22! I'm here to help you with any issues or questions you have. Let's work together to solve this.

I found a similar open issue regarding support for object storage in the Jaeger Helm charts: Feature: Support Jaeger-v2 Helm Chart, which is still open and was last updated on October 22, 2024 ^[1].

Currently, the Jaeger Helm chart supports storage backends like cassandra, elasticsearch, kafka (only as a buffer), grpc-plugin, badger, and memory ^[2]. There isn't direct support for object storage like AWS S3 or Azure Blob Storage in the Helm chart.

However, you can use the grpc-plugin storage type to integrate with third-party storage solutions. This involves setting up a gRPC-based plugin that can interact with your desired object storage. You would need to configure the Helm chart to use grpc-plugin and provide the necessary plugin configuration ^[3].

To continue talking to Dosu, mention @dosu.

[jkowall]

Hello, @asakth22. You would not use the helm chart to manage storage if you want to setup more complex tiering. In fact not even the operator would support this setup and you'd need a bigger cluster to manage it. Depending on your setup you might want hot/warm or hot/frozen with searchable snapshots. You can read more about different tiering and tradeoffs here in the elastic docs : https://www.elastic.co/guide/en/elasticsearch/reference/current/data-tiers.html

You would need to disable the Jaeger index management features and use ILM which is part of ElasticSearch : https://www.elastic.co/guide/en/elasticsearch/reference/current/index-lifecycle-management.html

You would have to rotate the indexes this way and also move from hot to warm/frozen depending on your requirements. This is a lot more work as you can tell, so it might not be worth it.

Maybe to ask another question why do you want to use object storage? Is if for retention or for cost?

You can likely solve both problems by looking at something like ClickHouse which is unsupported officially, but we hope to change that soon. I hope this helps.

[asakth22]

Thanks for your response. Yes, it's for cost and also for retention as we recreate the cluster and other resources every day.
The plan was to separate the storage part so the data can be backed up.
But I'll look for other feasible solution.