|
Title: Mandatory fees in 0.8 (and not in 0.7 it seems) Post by: evilpete on April 05, 2013, 06:11:59 AM I'm getting a lot of log noise about this in my logs:
ERROR: CTxMemPool::accept() : not enough fees 578a562743458c128144097996fc6a62e65bf6f8a7803bf5ca29325274429ad3, 0 < 10000 And looked up the transaction: https://blockchain.info/tx/578a562743458c128144097996fc6a62e65bf6f8a7803bf5ca29325274429ad3 It's been in limbo for 54+ hours now. Yes, it is a 0-fee transaction but it's small - 257 bytes. I was under the impression that fees weren't mandatory for this sort of thing. Sure, they'd be slow to confirm, but the backlog right now doesn't look too huge. The odd thing is the change address is sent back to itself. Presumably this is the work of something like the coin-control patches or another client? If I had to guess, I'd wager the person tried to send 0.005 btc to somebody, followed by 10btc to somebody else and is wondering why it's all in limbo.. The first one has been stuck for 54 hours, the second (10 btc) for 10-ish hours so far. Why do so many of the peers I'm connected to seem to think this is fine, but my 0.8.1's reject it out of hand? What's going to happen to it? Some miner with ancient software will eventually pick it up? The sender will have to purge the unconfirmable tx from their wallet? If so, that doesn't make for a great user experience.. Edit: https://en.bitcoin.it/wiki/Transaction_fees Edit: There was speculation that blockchain.info might be retransmitting these. I could be wrong about this coming from multiple clients, it might just be one since bitcoind doesn't log tx origins from which peers. Title: Re: Mandatory fees in 0.8 (and not in 0.7 it seems) Post by: Mike Hearn on April 05, 2013, 10:07:54 AM It can result from people modifying the default fee in bitcoinj clients. The small output would trigger a required fee, I think.
|