[Bug]: spendable balance 0uatom is smaller than 1uatom: insufficient funds with gas used

[Ekzer]

Is there an existing issue for this?

• I have searched the existing issues

What happened?

Hi,
version
For some reasons, since switch to v0.50.8-lsm, when I try to simulate send tx with full balance, I always get now that error :

\"spendable balance 0uatom is smaller than 1uatom: insufficient funds with gas used: '63581'

However, I didn't change anything regarding simulation.
I simulate with fee=0 in reference to that previous issue #17266
and then calculate the right amount to withdraw including fees.
This works since the beginning

Cosmos SDK Version

v0.50.8-lsm

How to reproduce?

It only fails on testnet right now. Mainnet is working still

You can try with these inputs :

{
    "tx_bytes": "CpMBCpABChwvY29zbW9zLmJhbmsudjFiZXRhMS5Nc2dTZW5kEnAKLWNvc21vczE2ZHF2bDg2a2M1bHI5cThyNWUwY3lsbjRnYXQ0NWg5enFocHR4MxItY29zbW9zMTlkczI2Nnp2NHdlM2x3ZjN3NzJrZHJoemV1cDhwZnlsNmN2MDJoGhAKBXVhdG9tEgc3NDM3ODYyElMKTwpGCh8vY29zbW9zLmNyeXB0by5zZWNwMjU2azEuUHViS2V5EiMKIQPd/rryoq14vd0oDxGpzGRzrNYfYdb7XkCk0BlrG6r2OhICCgAYlBgSABoA",
    "mode": "BROADCAST_MODE_SYNC"
}

Want to withdraw the full balance of this one for example https://www.mintscan.io/cosmoshub-testnet/address/cosmos16dqvl86kc5lr9q8r5e0cyln4gat45h9zqhptx3/
but you can pick any random address with some tokens, this results the same

EDIT :
After some more investigations, it works if you subtract 1 uatom to the amount...

[facundomedica]

Hey, as you correctly pointed out this is a feature we don't have yet. I understand that the simulation of tx would succeed on the previous version, but did the actual tx execution suceed? That is with the full amount of the wallet.

What might have happened is that now the simulation is "more truer" than before, so it fails just like it would fail on execution (I haven't found it in the changelog, so it's just a guess).

I think we should close this issue and track it with #17266

[Ekzer]

Hi @facundomedica , no this is not what I mean. The tx simulation fails now since v0.50.x
As I explained, I could simulate with full amount and fee = 0 without any issues. Now, it's not.
And it seems to pass when I substract just 1 uatom to the full amount, which sounds very strange.

[Ekzer]

The issue #17266 will allow estimation with full amount and fee > 0

[Ekzer]

From what I see, this error can only come from 2 places :

cosmos-sdk/x/accounts/defaults/lockup/lockup.go

Lines 564 to 574 in a020729

	if _, hasNeg := spendable.SafeSub(coin); hasNeg {
	if len(spendable) == 0 {
	spendable = sdk.Coins{sdk.NewCoin(coin.Denom, math.ZeroInt())}
	}
	return errorsmod.Wrapf(
	sdkerrors.ErrInsufficientFunds,
	"spendable balance %s is smaller than %s",
	spendable, coin,
	)
	}

or

cosmos-sdk/x/bank/keeper/send.go

Line 252 in a020729

"spendable balance %s is smaller than %s",

Same logic.

That means the spendable amount is somehow equal to 0 and coin equals 1 🤔

		if _, hasNeg := spendable.SafeSub(coin); hasNeg {
			if len(spendable) == 0 {
				spendable = sdk.Coins{sdk.Coin{Denom: coin.Denom, Amount: math.ZeroInt()}}
			}

[facundomedica]

Can you provide a practical example? Like the step by step you do to get this error, CLI commands or API requests you do. I wasn't able to replicate from CLI, I can do --dry-run and also tx simulate.

[Ekzer]

Hey @facundomedica :)
It's through REST Api.
You can test it once again with my example above.

curl --location 'http://$hostname/cosmos/tx/v1beta1/simulate' \
--header 'Content-Type: application/json' \
-XPOST
--data '{
    "tx_bytes": "CpMBCpABChwvY29zbW9zLmJhbmsudjFiZXRhMS5Nc2dTZW5kEnAKLWNvc21vczE2ZHF2bDg2a2M1bHI5cThyNWUwY3lsbjRnYXQ0NWg5enFocHR4MxItY29zbW9zMTlkczI2Nnp2NHdlM2x3ZjN3NzJrZHJoemV1cDhwZnlsNmN2MDJoGhAKBXVhdG9tEgc3NDE3NjM0ElMKTwpGCh8vY29zbW9zLmNyeXB0by5zZWNwMjU2azEuUHViS2V5EiMKIQPd/rryoq14vd0oDxGpzGRzrNYfYdb7XkCk0BlrG6r2OhICCgAYoBgSABoA",
    "mode": "BROADCAST_MODE_SYNC"
}'````

[facundomedica]

Hey @Ekzer didn't see your last reply, where are you getting those tx bytes?

[Ekzer]

Hey @facundomedica ,
Through the JS SDK.
Something like that :

const uri = '/cosmos/tx/v1beta1/simulate';
      const tx = Tx.fromPartial({
        authInfo: AuthInfo.fromPartial({
          fee: Fee.fromPartial({}),
          signerInfos: [
            {
              publicKey: encodePubKey(signer),
              sequence: BigInt(sequence),
              modeInfo: { single: { mode: SignMode.SIGN_MODE_UNSPECIFIED } },
            },
          ],
        }),
        body: TxBody.fromPartial({
          messages: messages.map((message) => this.registry.encodeAsAny(message))
          memo,
        }),
        signatures: [new Uint8Array()],
      });
      const request = {
        tx_bytes: base64FromBytes(Tx.encode(tx).finish()),
        mode: 'BROADCAST_MODE_SYNC',
      };

this.registry

comes from something like that :

import { Registry } from '@cosmjs/proto-signing';
import { defaultRegistryTypes } from '@cosmjs/stargate';
import { MsgSetOrchestratorAddresses } from '@injectivelabs/core-proto-ts/cjs/injective/peggy/v1/msgs';
import { MsgUnjail } from 'cosmjs-types/cosmos/slashing/v1beta1/tx';

export function getRegistry(): Registry {
  const registry = new Registry(defaultRegistryTypes);
  registry.register('/injective.peggy.v1.MsgSetOrchestratorAddresses', MsgSetOrchestratorAddresses);
  registry.register('/cosmos.slashing.v1beta1.MsgUnjail', MsgUnjail);
  return registry;
}

This works like a gem since day 1

[facundomedica]

Hey @Ekzer, also provide the message you are sending, as what you provided isn't enough for me to replicate.

EDIT: to be clear, we need to be able to replicate the issue in order to find a fix. So please be as detailed as possible.

[Ekzer]

Hey @facundomedica , check your Github inbox :) I invited you to the test repo : https://github.com/Ekzer/cosmjs-issue-21072

[Ekzer]

Btw, got the same in mainnet

{
    "code": 2,
    "message": "failed to execute message; message index: 0: spendable balance 9999uatom is smaller than 10000uatom: insufficient funds [cosmos/[email protected]/x/bank/keeper/send.go:278] with gas used: '87635'",
    "details": []
}

[Ekzer]

Hey @facundomedica, were you able to reproduce it ? :)

[Ekzer]

It looks like it has been resolved now 🤔 Can't reproduce it anymore 🎉
Has anyone changed something since?

[julienrbrt]

We weren't able to replicate the issue, but good to know that the problem was solved for you.