maker failed PLG20 swap saying that tx was not found on RPC node

[cipig]

maker failed a PLG20 swap because tx sent by taker was not found on the RPC node

the relevant part in maker's swap json:

      {
         "event" : {
            "type" : "TakerPaymentWaitConfirmStarted"
         },
         "timestamp" : 1674500829257
      },
      {
         "timestamp" : 1674500845249,
         "event" : {
            "type" : "TakerPaymentValidateFailed",
            "data" : {
               "error" : "maker_swap:970] !taker_coin.validate_taker_payment: eth:3457] Didn't find provided tx 0x74e4e262f31906d64b8733245062ace84250638100742e42d303214a3721d92c on ETH node"
            }
         }
      },

the console on maker showed this while it failed the swap:

23 19:07:09, maker_swap:917] INFO Taker payment tx 74e4e262f31906d64b8733245062ace84250638100742e42d303214a3721d92c
+--- 23 19:07:11 -------
| (0:49) [swap uuid=456d773b-dcec-49df-bdb2-587518ab7c1d] Taker payment wait confirm started...
| (0:01/103:10) [OCEAN-PLG20] Waiting for confirmations…
· 2023-01-23 19:07:24 +0000 [OCEAN-PLG20] Waiting for confirmations… Confirmed.
+--- 23 19:07:25 -------
| (1:03) [swap uuid=456d773b-dcec-49df-bdb2-587518ab7c1d] Taker payment wait confirm started...
+--- 23 19:07:33 -------
| (1:12) [swap uuid=456d773b-dcec-49df-bdb2-587518ab7c1d] Maker payment refund started...

the tx in question is this one: https://polygonscan.com/tx/0x74e4e262f31906d64b8733245062ace84250638100742e42d303214a3721d92c
it was mined just fine

maker was using https://polygon-rpc.com RPC node while taker was using the ones defined in https://github.com/KomodoPlatform/coins/blob/master/ethereum/MATIC, which are more

it's a bit strange that maker said that he found the tx ([OCEAN-PLG20] Waiting for confirmations… Confirmed) but failed the swap

both maker and taker were using the same mm2 from dev branch: http://195.201.0.6/dev/mm2-7201146b3-Linux-CI.zip

what can be the cause? sync issues between the different RPC nodes that taker was using? if so, can we make a change so that mm2 tries a second time couple seconds later?

[artemii235]

We use eth_getTransactionReceipt RPC in wait_for_confirmations and eth_getTransactionByHash in validate_payment. It seems that RPC node(s) might have a race condition when they return the receipt, but not the transaction itself yet. It's worth to call eth_getTransactionByHash in wait_for_confirmations too to prevent the problem.

Should be easy, so it looks like a good test task.

[cipig]

i had a very similar issue and it sounds like the above solution will solve this problem too
so i was trying to swap RNDR-PLG20 for SYS as a taker
and taker "said" this during takerpayment (RNDR-PLG20)

11 07:51:50, taker_swap:1319] INFO After wait confirm
+--- 11 07:51:51 -------
| (8:28) [swap uuid=102ff55d-25df-4d8b-9a20-b5ba05b3370c] Maker payment validated and confirmed...
| (0:00) [sign-and-send] get_gas_price…
· 2023-02-11 07:51:52 +0000 [sign-and-send] get_addr_nonce…
· 2023-02-11 07:51:55 +0000 [sign-and-send] get_addr_nonce…
11 07:51:55, taker_swap:1439] INFO Taker payment tx hash 31c91001eb35c9f80e1f56ab416c4425fea87b8d59663f27c5e140361f29c6a9
+--- 11 07:52:00 -------
| (8:36) [swap uuid=102ff55d-25df-4d8b-9a20-b5ba05b3370c] Taker payment sent...
| (0:05/121:27) [RNDR-PLG20] Waiting for confirmations…

just that the tx https://polygonscan.com/tx/0x31c91001eb35c9f80e1f56ab416c4425fea87b8d59663f27c5e140361f29c6a9 is nowhere to be found
a second try at a later time completed the exact same swap successfully
maker and taker were using latest mm2 from dev: 2.1.9934_dev_60337cf46_Linux_CI

[shamardy]

Although this is a good issue for a test task, it's better to fix this bug as soon as possible since it leads to failed swaps. Will work on it this week.

[shamardy]

@cipig Did you encounter these errors again? If not, I guess this issue can be closed.

[cipig]

saw it once in the last time https://dexapi.cipig.net/public/error.php?uuid=f691a5f1-e639-47c2-8ec5-cd7bdb224d17
but idk the circumstances, so closing