[QOS][SOLANA] Implement Solana Program Checks

[fredteumer]

Objective

Morse service is described as supporting several SPLs. These checks should ensure that only nodes that support these SPLs receive relays.

Origin Document

See Grove Docs: https://docs.grove.city/solana-api/endpoints/solana

Goals

• Ensure support for required SPLs

Deliverables

Implement QoS checks that ensure support for the following SPLs on solana service.

• Solana Kin Program
• Token Program with Indexing
• Associated Token Account Program

Non-goals / Non-deliverables

• Archival checks on Solana
• Sync check on Solana
• Any other QoS checks

General deliverables

• Comments: Add/update TODOs and comments alongside the source code so it is easier to follow.
• Testing: Add new tests (unit and/or E2E) to the test suite.
• Makefile: Add new targets to the Makefile to make the new functionality easier to use.
• Documentation: Update architectural or development READMEs; use mermaid diagrams where appropriate.

---

Creator: @fredteumer

[adshmh]

We need to refine these requirements to clarify the exact process of establishing QoS on a Solana endpoint:

• Umbrella GH issue for Solana QoS.
• Separate GH issue for each item: e.g. Token Program with Indexing.
• High-level business logic on each item: e.g. what observations do we need from any Solana endpoint to establish it supports the Token Porgram.

[Olshansk]

tl;dr @adshmh Do you have this handeled or do we need to clarify something?

---

@adshmh I read your messages above but it's not really clear to me what the ask/problem/next step here.

• Do you know exactly what needs to be done and are on it?
• Do you need clarification and/or direction?
• Are there open questions we need to align on?

I realize these are super "meta level" questions - but am looking for clarity to do something actionable.

[fredteumer]

Will await commentary before taking any action. Let me know if you want sub-issues or separate PRs, whatever works for tracking this.

[adshmh]

tl;dr @adshmh Do you have this handeled or do we need to clarify something?

@adshmh I read your messages above but it's not really clear to me what the ask/problem/next step here.

* Do you know exactly what needs to be done and are on it?

* Do you need clarification and/or direction?

* Are there open questions we need to align on?

I realize these are super "meta level" questions - but am looking for clarity to do something actionable.

In order to implement the required QoS features, we will need a high-level but still detailed description of the requirements.
More specifically, the answer to this question:

Given a set of Solana endpoints, what data points do we need to collect on each endpoint to establish its support for a specific program?

e.g. the answer to this question for EVM archival checks is something like the following:

1. Wait for the perceived block number to be established (via observations)
2. Calculate a random block number within e.g. 128 blocks of the perceived current block number.
3. Send `eth_getBalance` requests to each endpoint, with the configured contract and the calculated block number.
4. Take the mode value, i.e. the balance reported the most by the endpoints, as the correct value.
5. Disqualify an endpoint if its reported balance does not match the above.

[Olshansk]

@adshmh With the help of perplexity to build some context, and ChatGPT's o3 model, I have come up with the description below.

Link to ChatGPT conversation for reference.

Disclaimer: I don't know if this is correct. or if it works, but a spot check of a few things (addresses, functions) looks good. From an intuition perspective, it makes sense to me as well.

---

Program IDs (configurable)

[solana_qos.programs]
# Kin Program
KIN   = "kinXdEcpDQeHPEuQnqmUgtYykqKGVFq6CeVX5iAHJq6"
# SPL Token Program
TOKEN = "TokenkegQfeZyiNwAJbNbGKPFXCWuBvf9Ss623VQ5DA"
# SPL Associated-Token-Account Program
ATA   = "ATokenGPvbdGVxr1b2hvZbsiqW5xWH25efTNsLJA8knL"

---

Per-program Check Matrix

Check type	RPC call & key param	Pass condition	Why it matters (clearer)
Existence	getAccountInfo(programId)	account.executable == true	Verifies the node has the program binary deployed
Indexed read	getProgramAccounts(programId, {limit:1})	Returns ≥ 1 account within 1 s	Confirms the node keeps an index for program accounts
Modal accuracy	Recent finalized slot → getAccountInfo(referenceAccount)	Value matches mode across peers	Detects nodes with stale or wrong state
(Optional) simulate	simulateTransaction with no-op instruction	err == null	Proves the runtime can actually execute the program

Equal weighting: fail any check for any required program → endpoint is disqualified for this cycle.

---

Probe Cycle Flow

flowchart TD
    Start([Begin probe cycle])
    Start --> EP{For each endpoint}
    EP --> PG{For each configured program}
    PG --> A[Existence check]
    A --> B[Index test]
    B --> C[Modal accuracy]
    C -->|All pass| OK[Route traffic]
    C -->|Any fail| KO[Disqualify endpoint]
    OK --> Next((Next endpoint))
    KO --> Next
    Next -->|loop| EP

Loading

[Olshansk]

@fredteumer If you have feedback on the approach above, please share it. Not something I've looked into prior to today, so it'll be helpful.

[fredteumer]

Agree with above approach and based on my intermediate knowledge of the idiosyncrasies of Solana RPC, seems functionally correct.

[Olshansk]

@adshmh @fredteumer I'll also reach out to some people in the solana community and/or those who use solana endpoints to get some feedback.

If we get something in place though, it'll be easier to poke holes at.