BitcoinArmory
BitcoinArmory copied to clipboard
Reconnect to Bitcoin-QT logic flawed
It appears that there is a bug in Armory when it is trying to re-connect to Bitcoin-QT. I am running btcd instead of Bitcoin-QT. At first Armory shows that it is connected, but eventually, it disconnects (I assume this is because the btcd differs from Bitcoin-QT in some way). Every time Armory reconnects it appears to attempts an additional socket. Eventually, it tries to connect so many times that it uses up all of btcd's inbound connections.
This should be easy to reproduce, but let me know if you need logs.
There shouldn't be any difference between Bitcoin-Qt and bitcoind. Armory only uses the P2P interface, too, which is even more "standardized" than the RPC inteface (though they should both be identical in both versions).
Either way, I don't have much control over that. I'm using the "ReconnectingClientFactory" which is part of python-twisted, to maintain the connection between Armory and Bitcoin-Qt/bitcoind. It automatically reopens the connection when it's lost, and probably does so on a new socket everytime. Perhaps there's a way to tell it to reuse the same socket...? Unfortunately, I'm not too familiar with that side of things.
I don't know if this helps, but once its messed up, it appears that just launching QT doesn't fix it. When it is doing the disconnect/reconnect thing, the balloon from the system tray just flickers.
@etotheipi Just a remark, he mentions btcd, which is not bitcoind. It looks like you assumed he meant bitcoind.
I'm looking for the possibility to use Armory with btcd. Does Armory still need to access the block files bitcoind produces? Or can it already work using the RPC api and txindex=1?