[tempo] - Persistence for compactor is not aligned with other components

[Rohlik]

Hi there 👋🏼,
compactor within Tempo seems to be slightly off in setting persistence compared to other components like Mimir or Loki 🧐. Please take a look at the related code snippets below.
Mimir: https://github.com/grafana/mimir/blob/bec189c4ba11111a251c3b48253704736761ad23/operations/helm/charts/mimir-distributed/values.yaml#L2282-L2320
Loki: https://github.com/grafana/loki/blob/226e9f101aab2faf32b2d6b0c93f6e5832782c25/production/helm/loki/values.yaml#L2430-L2459

There is no such setting for Tempo for some reason 🤔, and current behavior can potentially lead to a situation when Tempo fulfills the worker's disk capacity.

Is this desired 🤔? If not, can you please add this option to the Tempo Helm chart 🙏🏼?

[diegocejasprieto]

I have the same question... why does compactor not have a "persistence" section or something? Is there any reason why it shouldn't persist any data?
I know that "ingester" needs to have any kind of persistance before uploading any data to a backend storage (i.e S3) in case of any pod restart... but I cannot understand why compactor does not include any mechanism for doing so.