Discussion on: [WIP] Testnet RPC not reliable

View post

Daniel Liu • Oct 18 '23 • Edited on Oct 18

Could you please share the below files?

docker-compose yaml file
more log messages
subgraph project

So I can reproduce your problem.

I don't have a docker compose setup to share since this is happening in production. This also isn't happening because of a specific subgraph but during initial network setup in the graph node.

Here are more logs: gist.github.com/paymog/33659ee46f0...

Here's how you can manually reproduce this:

curl -X POST -H "Content-Type: application/json" -d '{"method": "eth_getBlockByHash", "params": ["0xd28bc799751815a0923d673d97dc8ca932e32c41e1d1f8ec08956106bd981018", false], "id": 1, "jsonrpc": "2.0"}' https://earpc.apothem.network which contains a single transaction: 0xc1b52110f52120b4faf9e8502d568fb6dc88e90b0acd9f48d0af443782b9f55b
curl -X POST -H "Content-Type: application/json" -d '{"method": "eth_getTransactionReceipt", "params": ["0xc1b52110f52120b4faf9e8502d568fb6dc88e90b0acd9f48d0af443782b9f55b"], "id": 1, "jsonrpc": "2.0"}' https://earpc.apothem.network which apparently belongs to block 0x037d2d9f0e9e5c38ee9fb0c6721f5873cee7984bfe3bca2f91974daadcebbfd7

Note that the two block hashes, 0xd28bc799751815a0923d673d97dc8ca932e32c41e1d1f8ec08956106bd981018 and 0x037d2d9f0e9e5c38ee9fb0c6721f5873cee7984bfe3bca2f91974daadcebbfd7, differ and match the log statements which say

graph-node Oct 20 12:12:40.288 INFO receipt block mismatch, receipt_tx_hash: 0xc1b5…f55b, tx_hash: 0xc1b5…f55b, block_hash: 0xd28b…1018, receipt_block_hash: 0x037d…bfd7, provider: xdc-testnet, component: EthereumPollingBlockIngestor

Daniel Liu • Oct 20 '23

Please set ETHEREUM_REORG_THRESHOLD to 1. You can refer to github.com/XinFinOrg/XDC-Graph-Nod....

Paymahn Moghadasian • Oct 23 '23

That seems like a bandaid solution that will end up with incorrect data long term. From the graph node docs

ETHEREUM_REORG_THRESHOLD: Maximum expected reorg size, if a larger reorg happens, subgraphs might process inconsistent data. Defaults to 250.

Note that we're not seeing this against the XDC mainnet, only the testnet. Are the RPC calls I provided above expected and correct behavior?

Daniel Liu • Oct 23 '23

I'm not sure whether the mainnet has this issue. This issue will affect some software. But It's a very big project to solve it.

Daniel Liu • Oct 24 '23 • Edited on Oct 24

I'll write a script to check the duplicate transaction hashes on mainnet when I'm free.

Daniel Liu • Nov 14 '23

The mainnet has same transaction hash bug also, such as both of blocks 39185599 and 39185611 has transaction hash 0x000f94c5677e68504bc8daabdb051d3f1e51fd626b59d9533f5fc55f656f6fd1. There is a fix already: github.com/XinFinOrg/XDPoSChain/pu.... Please wait for some time.

Daniel Liu • Oct 20 '23 • Edited on Oct 20

In your case, both of blocks 0x354ce39 and 0x354ce45 contain the transaction hash 0xc1b52110f52120b4faf9e8502d568fb6dc88e90b0acd9f48d0af443782b9f55b. This is reported in github.com/XinFinOrg/XDPoSChain/pu.... Maybe this is a problem or bug. I will study your log messages later. Would you please share below files?

graph-node configuration
subgraph source codes

curl -s -X POST -H "Content-Type: application/json" https://earpc.apothem.network -d '
{
  "jsonrpc": "2.0",
  "id": 4003,
  "method": "eth_getBlockByNumber",
  "params": [
    "0x354ce39",
    true
  ]
}' | jq

curl -s -X POST -H "Content-Type: application/json" https://earpc.apothem.network -d '
{
  "jsonrpc": "2.0",
  "id": 4004,
  "method": "eth_getBlockByNumber",
  "params": [
    "0x354ce45",
    true
  ]
}' | jq

Developers Forum for XinFin XDC Network

Developers Forum for XinFin XDC Network is a community of 2,909 amazing users

Discussion on: [WIP] Testnet RPC not reliable