`eth_getTransactionByHash` returns `null` for older transactions. Why? When?

Question

I was looking into an issue with RPC calls with eth_getTransactionByHash and recognized for older transactions it returns null for geth full nodes:

Tx 511 days ago: 0x9f15fa4b2fe31afa2953331dfb58662422c6eb6b4fb3a739ae6f40c469979695.

Requesting own local full node:

kuzdogan@kuzdogan-UX331UN:~$ curl --data '{"method":"eth_getTransactionByHash","params":["0x9f15fa4b2fe31afa2953331dfb58662422c6eb6b4fb3a739ae6f40c469979695"],"id":1,"jsonrpc":"2.0"}' -H "Content-Type: application/json" -X POST http://localhost:8545
{"jsonrpc":"2.0","id":1,"result":null}

Requesting an Infura node:

curl --data '{"method":"eth_getTransactionByHash","params":["0x9f15fa4b2fe31afa2953331dfb58662422c6eb6b4fb3a739ae6f40c469979695"],"id":1,"jsonrpc":"2.0"}' -H "Content-Type: application/json" -X POST https://mainnet.infura.io/v3/XXXXXXX
{"jsonrpc":"2.0","id":1,"result":{"blockHash":"0x15fac326a1dabb09acbe5727fb670f767733e9cb490e6329e3970990b67e9810","blockNumber":"0xa6116a","from":"0xdc32476dc0d5442836439924697166185f4d9c6b","gas":"0x6270","gasPrice":"0x2cb4178000","hash":"0x9f15fa4b2fe31afa2953331dfb58662422c6eb6b4fb3a739ae6f40c469979695","input":"0x","nonce":"0x21d","r":"0x4f8e7b7ab538f05cb39a0ec1031a0c40961afa067aa3cba1c23621263c435071","s":"0x32b0b5f4c014e2cea9724bc823da3a98e8341b9f29942388fe7aa07ce439fc4a","to":"0xdcb2fb145e86f31245efa748650671fcac39232f","transactionIndex":"0xa0","type":"0x0","v":"0x26","value":"0x69789fbbc4f80000"}}

Noting it is possible to retrieve the same tx via eth_getTransactionByBlockNumberAndIndex. And for recent transactions, eth_getTransactionByHash works.

I guess the full nodes stop indexing transactions when they are old. (Moved to "ancient"? Pruned?).

So my questions are:

• Is it because the transactions are removed from "index" in a geth full node?
• If so, when is this done? 1 year? 6 months? N blocks?
• How could a transaction be retrieved in this case by only knowing the transaction hash?

Answer 1

I learned this is due to the following option in geth:

--txlookuplimit value               Number of recent blocks to maintain transactions index for (default = about one year, 0 = entire chain) (default: 2350000)

So yeah it is by default set to ~1year and my local node should have this value where Infura nodes probably have the entire chain set.