[v0.11 feature] Per-chain Relays To Token Multiplier #1580

msmania · 2023-09-07T00:52:05Z

Description

This patch introduces a new network parameter pos/RelaysToTokensMultiplierMap so that we can set a custom RTTM per chain. The existing parameter pos/RelaysToTokensMultiplier serves the default RTTM if a custom RTTM is not set.

This change is consensus-breaking. The new behavior is put behind a new feature key RTTM2. The new parameter is kept unavailable until activation.

The patch also contains two tests:

Test_RTTM2_ChangeParamValue to make sure the new parameter is kept unavailable until activating the RTTM2 feature key.
TestKeeper_RewardForRelaysPerChain to make sure the amount of relay rewards are calculated with a custom RTTM in the new parameter if it's set.

Summary generated by Reviewpad on 07 Sep 23 00:52 UTC

This pull request contains the following changes:

The addition of a new test function called TestKeeper_RewardForRelaysPerChain. This test function verifies the rewards generated for different scenarios related to relays per chain.
Modifications to the file "expectedKeepers.go" were made. A new function called "RewardForRelaysPerChain" was added, which calculates and returns a value of type sdk.BigInt.
Changes were made to the file "module.go" related to activating additional parameters and setting a minimum signed per window value.
The file "keeper_test.go" includes the addition of a new test function called "Test_RTTM2_ChangeParamValue", which tests the change of a parameter value after a certain activation height.
A new file "proof.go" was modified to enhance the functionality and reliability of the ExecuteProof function in handling relay proofs.
The file "param.go" now includes additional tracking of parameters introduced after genesis and skips writing them to the database until the corresponding feature activation height is reached.
The file "nodes.go" includes changes to the AwardCoinsForRelays function, enabling it to award coins for relays completed per chain.
Modifications were made to the PosKeeper interface in the file "expectedKeepers.go", adding a new function called RewardForRelaysPerChain.
The file "service_test.go" includes changes to the RewardForRelays method in the MockPosKeeper struct, allowing coins to be awarded for relays completed per chain.
The file "abci.go" in the x/nodes/keeper package includes import statement additions and changes to variable names.
The go.sum file shows updates to dependencies related to the "github.com/pokt-network/tendermint" package.

These changes enhance functionality, introduce new features, and improve reliability in various parts of the codebase.

This patch imports pokt-network/tendermint@5931015.

The parameter `pos/MinSignedPerWindow` defines a proportion, but the function `MinSignedPerWindow` returns the number of blocks, calculated as `pos/MinSignedPerWindow * pos/SignedBlocksWindow`. This is confusing. To resolve this discrepancy, we had "custom logic" to convert the return value back to the parameter value. This is unnecessary calculation. There was indeed a bug in `PosKeeper.GetParams` that returned the return value from `k.MinSignedPerWindow(ctx)` without the "custom logic". As a result, the value of `min_signed_per_window` in the response from /v1/query/nodeparams was always wrong. To remove this confusing design, this patch does: - Rename the function `MinSignedPerWindow` to `MinBlocksSignedPerWindow` - Add a new `MinSignedPerWindow` function returning the exact parameter value in decimal as the other parameter getter functions do

The function `SetParamSet` writes parameters to application db, where we have a special logic to prevent parameters introduced after genesis from being written in blocks before those parameters are introduced. However, using `IsOnUpgradeHeight` in this logic is wrong because all post-genesis parameters we have are activated by a feature name, not when the protocol version is upgraded. As a result, we had never exercised this logic. This didn't cause any real problem yet because we never call `SetParamSet` for post-genesis parameters before they're introduced. We should do the right thing before this bug causes problems in the future.

POKT-Discourse · 2023-09-08T17:17:13Z

This pull request has been mentioned on Pocket Network Forum. There might be relevant details there:

https://forum.pokt.network/t/unleashing-the-potential-of-pocket/4720/1

POKT-Discourse · 2023-10-13T22:38:14Z

This pull request has been mentioned on Pocket Network Forum. There might be relevant details there:

https://forum.pokt.network/t/unleashing-the-potential-of-pocket-as-universal-rpc-provider/4797/1

Olshansk

Overall: LGTM! Great change. Simple & powerful!

Left some small NITs, comments and questions but nothing important. Just want to go over it before merging
Make sure to update the documentation (when we're closer to launch) on how to execute the tx commands with the marshalled json map
The ONLY concern I have is related to the cleanup you did related to the MinSignedPerWindow. I don't know why we had this before and wanted to understand your confidence level on it?

// custom logic for minSignedPerWindow
		params.MinSignedPerWindow = params.MinSignedPerWindow.QuoInt64(params.SignedBlocksWindow)

app/tx_test.go

x/gov/types/expectedKeepers.go

Olshansk · 2023-11-29T03:48:53Z

x/nodes/keeper/abci.go

@@ -25,7 +26,7 @@ func BeginBlocker(ctx sdk.Ctx, req abci.RequestBeginBlock, k Keeper) {
 	// store whether or not they have actually signed it and slash/unstake any
 	// which have missed too many blocks in a row (downtime slashing)
 	signedBlocksWindow := k.SignedBlocksWindow(ctx)
-	minSignedPerWindow := k.MinSignedPerWindow(ctx)
+	minSignedPerWindow := k.MinBlocksSignedPerWindow(ctx)


In x/nodes/keeper/params.go, we use MinSignedPerWindow and here we use MinBlocksSignedPerWindo.

Just confirming that this is intended?

Yes. As described in the Part2 commit's message, the function k.MinSignedPerWindow is renamed to k.MinBlocksSignedPerWindow. This function returns int64, the number of blocks.

The parameter value MinSignedPerWindow (not a function), which is BigDec and the current value is 0.6, is used in only one place, GetParams in x/nodes/keeper/params.go.

Olshansk · 2023-11-29T03:50:06Z

x/nodes/keeper/keeper.go

@@ -107,8 +107,6 @@ func (k Keeper) ConvertState(ctx sdk.Ctx) {
 		panic(err)
 	}
 	k.Cdc.SetUpgradeOverride(true)
-	// custom logic for minSignedPerWindow


I'm not familiar with this whole manual update for MinSignedPerWindow so it's the only piece that's giving me a bit of "unease" but I don't know how to confirm/validate it.

Is your confidence level high on these changes?

The ONLY concern I have is related to the cleanup you did related to the MinSignedPerWindow. I don't know why we had this before and wanted to understand your confidence level on it?

For the details, please check the commit message here.

The reason why we have this "custom logic" is the param SignedBlocksWindow is a proportion (= 0.6 in the mainnet), while the function SignedBlocksWindow is expected to return the number of blocks in a window (= 6 in the mainnet). This custom logic converts the proportion into the number of blocks by multiplying SignedBlocksWindow. This patch reconciles this discrepancy under the name SignedBlocksWindow.

The behavior is clear and the regression risk is low.

x/nodes/keeper/reward_test.go

Olshansk · 2023-12-14T00:37:49Z

@msmania Ty for linking to the commit messages. I don't want this to become a "developer workflow" discussion since I know everyone has opinions/preferences on how to use git, github, IDEs, etc.

I'm sharing mine below not in expectation of others adopting it, but just so you have context/visibility into how I work and I will accommodate your workflow as well.

I read the GitHub PR description
I go through all the files 1 by 1 (see screenshot below), checking them once I'm done.
I use the "show changes since last review" when I do round 2, 3, etc...
In GitHub land (note: not git), I don't pay attention to commits at all (and don't put effort into my own commits), since we end up doing a squash and merge to main at the end.

msmania · 2023-12-15T08:22:28Z

Got it. I personally thought one gigantic commit would be hard to be reviewed, or to be analyzed after merge in the future, so I split a patch to several PartX commits, following a convention in the projects I used to work. Anyway, thank you for signing off! Let's merge this!

POKT-Discourse · 2024-01-31T19:00:10Z

This pull request has been mentioned on Pocket Network Forum. There might be relevant details there:

https://forum.pokt.network/t/rc-0-11-1-upgrade-and-hi/5012/1

This is a fix for a regression introduced by Per-chain RTTM (#1580). When we store a map-type parameter, we use `Codec.MarshalJSON` that doen't guarantee the order of fields in a map. This causes consensus failure because a different order of fields in the parameter store generates a different merkle tree (= AppHash). The proposed fix is to canonicalize the marshaled bytes by calling `MustSortJSON` before storing it in the tree.

This is a fix for a regression introduced by Per-chain RTTM (#1580). When we store a map-type parameter, we use `Codec.MarshalJSON` that doen't guarantee the order of fields in a map. This causes consensus failure because a different order of fields in the parameter store generates a different merkle tree (= AppHash), resulting in chainhalt. The proposed fix is to canonicalize the marshaled bytes by calling `MustSortJSON` before storing it in the tree. Changing `Subspace.Set` is enough and a change in ` Subspace.SetWithSubkey` won't be necessary because it's not used from anywhere, but we should keep consistency. reviewpad:summary

kinesis-agent added 4 commits September 2, 2023 16:18

Part 1 - Import the latest tendermint 0.33.8

d215121

This patch imports pokt-network/tendermint@5931015.

Part 4 - Per-chain RTTM

0bd97f7

reviewpad bot added large Pull request is large waiting-for-review labels Sep 7, 2023

msmania mentioned this pull request Sep 7, 2023

[v0.11 feature] Built-in Reward Share #1581

Merged

msmania self-assigned this Nov 14, 2023

msmania requested a review from Olshansk November 14, 2023 23:16

Olshansk requested changes Nov 29, 2023

View reviewed changes

Olshansk modified the milestones: Network Scalability, Network Cost Nov 29, 2023

Olshansk mentioned this pull request Dec 1, 2023

feat: enforce max chains during new session #1582

Merged

address comments

78971e9

msmania requested a review from Olshansk December 6, 2023 00:52

Olshansk approved these changes Dec 14, 2023

View reviewed changes

msmania merged commit b48605e into staging Dec 15, 2023
4 of 5 checks passed

msmania deleted the tokikuch/RTTM2 branch December 15, 2023 08:23

Olshansk mentioned this pull request Apr 25, 2024

[Utility] Per-chain Relays To Token Multiplier pokt-network/poktroll#494

Closed

9 tasks

msmania mentioned this pull request Oct 1, 2024

Parameter JSON serialization must be deterministic #1614

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[v0.11 feature] Per-chain Relays To Token Multiplier #1580

[v0.11 feature] Per-chain Relays To Token Multiplier #1580

msmania commented Sep 7, 2023 •

edited by reviewpad bot

Loading

POKT-Discourse commented Sep 8, 2023

POKT-Discourse commented Oct 13, 2023

Olshansk left a comment

Olshansk Nov 29, 2023

msmania Dec 5, 2023

Olshansk Nov 29, 2023

msmania Dec 6, 2023 •

edited

Loading

Olshansk commented Dec 14, 2023

msmania commented Dec 15, 2023

POKT-Discourse commented Jan 31, 2024

[v0.11 feature] Per-chain Relays To Token Multiplier #1580

[v0.11 feature] Per-chain Relays To Token Multiplier #1580

Conversation

msmania commented Sep 7, 2023 • edited by reviewpad bot Loading

Description

Summary generated by Reviewpad on 07 Sep 23 00:52 UTC

POKT-Discourse commented Sep 8, 2023

POKT-Discourse commented Oct 13, 2023

Olshansk left a comment

Choose a reason for hiding this comment

Olshansk Nov 29, 2023

Choose a reason for hiding this comment

msmania Dec 5, 2023

Choose a reason for hiding this comment

Olshansk Nov 29, 2023

Choose a reason for hiding this comment

msmania Dec 6, 2023 • edited Loading

Choose a reason for hiding this comment

Olshansk commented Dec 14, 2023

msmania commented Dec 15, 2023

POKT-Discourse commented Jan 31, 2024

msmania commented Sep 7, 2023 •

edited by reviewpad bot

Loading

msmania Dec 6, 2023 •

edited

Loading