[DOCS] task/session/event data lifecycle

[fooius]

📋 Prerequisites

• I have searched the existing issues to avoid creating a duplicate
• By submitting this issue, you agree to follow our Code of Conduct

📝 Issue Description

I’m trying to understand the intended lifecycle of task, session, and event data in kagent.
By default the controller uses SQLite on a memory-backed emptyDir, which makes sense for quick demos or local experimentation. For longer-lived or HA setups, the documentation recommends moving to Postgres, which I agree with and plan to do.
What I haven’t been able to find, though, is any guidance on the expected lifecycle or retention of the task/session/event data itself. In our dev environment, the default 500Mi memory-backed emptyDir filled up fairly quickly under normal usage. As a temporary mitigation, I switched it to a 10Gi disk-backed emptyDir (still using SQLite for now), and I’m planning to migrate to Postgres later this week.
Before doing that, I’m trying to understand what a “reasonable” storage plan looks like:

• Is this data expected to grow indefinitely?
• Are there tables or categories of data (e.g. task events vs. task metadata) that are safe to prune?
• Is there any built-in retention or cleanup mechanism, or is retention expected to be handled externally by operators?

Any guidance on the intended data lifecycle, storage sizing expectations, or recommended retention strategies (especially when using Postgres) would be really helpful.

Thanks.

✅ Suggested Content or Change

No response

🙋 Willing to Contribute?

• I am willing to submit a PR for this documentation update