rfc19-user-bgp-status

[juan-malbeclabs]

This pull request introduces RFC 17, which proposes a new onchain mechanism to track the real BGP session status for each user, closing the gap between configuration and actual client connectivity. The RFC details new fields to be added to the User account, a new instruction for updating BGP status, and changes to telemetry collection and SDKs. This enhancement will enable more accurate diagnostics and user lifecycle management by making BGP session state observable onchain.

User account and onchain status tracking:

• Adds three new fields to the User struct: bgp_status (session state), last_bgp_up_at (last time session was Up), and last_bgp_reported_at (last time status changed), requiring a 17-byte account reallocation on first write.
• Introduces a new instruction, SetUserBGPStatus, allowing authorized metrics publishers to update BGP status for each user, with strict validation.

Telemetry and agent changes:

• Updates the telemetry collector to determine each user's BGP state after every collection tick and enqueue onchain status updates via the new instruction, using a background worker for reliability.

SDK and compatibility:

• Requires updates to SDKs (Go, TypeScript, Python) to handle the new fields, ensuring backward compatibility by treating missing fields as

[ben-dz]

I think overall DZDs should be using their telemetry key to write this data onchain. I don't think we want this as part of the onchain user, since I think we don't want to give DZDs the authority to modify user accounts.

instead, probably a separate onchain program (bgp-monitor?) that maintains this data. Eventually perhaps add a cross-program call to have the bgp-monitor call into serviceability to delete users when uploaded data shows stale sessions.

[juan-malbeclabs]

We would be creating an instruction that can be executed by the agent using the telemetry PK, which is only allowed to update this field in the user account. It does not have permission to modify any other fields. Additionally, it internally sets the slot values.

New instruction: SetUserBGPStatus (variant 94)

[ben-malbeclabs]

Approved from a conceptual perspective, would be good to have another dev confirm too.