[Bug]: Rollback will not re-execute the transaction.

[gofmanaa]

Is there an existing issue for this?

• I have searched the existing issues

What happened?

Rollback will not re-execute the transaction.

I found a problem with the rollback command.
The documentation about the rollback says that "the transactions in block n will be re-executed against the
application." but this does not happen.
Transactions with non-deterministic data are not re-executed and the blockchain does not produce new blocks.

Maybe I'm doing something wrong, please tell me how it should be, this question is very important to me

I want to know how such a problem can be solved.
What to do if it happens on prod?

Cosmos SDK Version

v0.47.2, v0.47.3

How to reproduce?

Steps to reproduction:
1 versions:
$ ignite version
Ignite CLI version: v0.27.1-dev
Ignite CLI build date: 2023-06-02T20:29:26Z
Ignite CLI source hash: 15cf9a60b8dfbed6d9ddb0cad7e890d2a5b60979
Ignite CLI config version: v1
Cosmos SDK version: v0.47.2
Your OS: linux
Your arch: amd64
Your Node.js version: v18.15.0
Your go version: go version go1.20.7 linux/amd64
Your uname -a: Linux archlinux 6.4.12-arch1-1 #1 SMP PREEMPT_DYNAMIC Thu, 24 Aug 2023 00:38:14 +0000 x86_64 GNU/Linux
Your cwd: /home/user
Is on Gitpod: false

2 Scaffold a new blockchain using ignite cli (https://docs.ignite.com/guide/blog/express)
3. Add non-determenistic data to message

func (k msgServer) CreatePost(goCtx context.Context, msg *types.MsgCreatePost) (*types.MsgCreatePostResponse, error) {
	ctx := sdk.UnwrapSDKContext(goCtx)

	var post = types.Post{
		Creator: msg.Creator,
		Title:   fmt.Sprintf("%d-%s", rand.Int(), msg.Title), #  <--add this random data
		Body:    msg.Body,
	}

5. Launch a blockchain on 3 nodes ()
   $ blogd testnet start
   acquiring test network lock
   preparing test network with chain-id "blog"
   +++++++++++++++++++++++++++++++++++++++++++++++++++++++++
   ++ THIS MNEMONIC IS FOR TESTING PURPOSES ONLY ++
   ++ DO NOT USE IN PRODUCTION ++
   +++++++++++++++++++++++++++++++++++++++++++++++++++++++++
   starting test network...
   started validator 0
   started validator 1
   started validator 2
   started test network at height: 1
   press the Enter Key to terminate

6. Broadcast the transaction. The message must contain random data.
   $ blogd tx blog create-post 'Hello, World!' 'This is a blog post' --from node0 --keyring-backend test --keyring-dir ./.testnets/blog/node0/simcli

The non-deterministic transaction stops the generation of new blocks.

7. Stop the blockchain and execute the rollback command
   $ blogd rollback --hard --home ./.testnets/blog/node0/simd
   $ blogd rollback --hard --home ./.testnets/blog/node1/simd
   $ blogd rollback --hard --home ./.testnets/blog/node2/simd
   Every time the command gives an error:
   Error: failed to rollback to version: no versions found while trying to load 6

8. Fix the message, removing random from the message.

func (k msgServer) CreatePost(goCtx context.Context, msg *types.MsgCreatePost) (*types.MsgCreatePostResponse, error) {
	ctx := sdk.UnwrapSDKContext(goCtx)

	var post = types.Post{
		Creator: msg.Creator,
		Title:   msg.Title, # <- fix message
		Body:    msg.Body,
	}

9. Rebuild the blockchain.
   $ ignite c build
10. Launch a blockchain on 3 nodes
    The blockchain changes the current height to n -1, but the transaction is not re-execute!

[facundomedica]

EDIT: replied without seeing that we re-implement this command in the SDK and that it's not only from Comet

[alexanderbez]

Error: failed to rollback to version: no versions found while trying to load 6 This seems problematic, no?

[gofmanaa]

Thanks, without a flag --hard, the node started working and new blocks started to be produced.

Description from CometBFT:

A state rollback is performed to recover from an incorrect application state transition,
when CometBFT has persisted an incorrect app hash and is thus unable to make
progress. Rollback overwrites a state at height n with the state at height n - 1.
The application should also roll back to height n - 1. If the --hard flag is not used, 
no blocks will be removed so upon restarting CometBFT the transactions in block n will be 
re-executed against the application. Using --hard will also remove block n. This can
be done multiple times.

Please tell me in what cases it is possible to use the "--hard" flag ?

[tac0turtle]

the error means that it could be that the node has pruned the older versions, can you share the pruning settings?