You aren’t using a reverse proxy, are you?
It’s a typical I/O error with nginx when load is getting too high
if i say i don’t know what a reverse proxy is, does that mean no?
I’m running both servers in one python directory (which is a bad idea because the logs get confusing, but it’s what I’m doing). I had both server scripts running for 27 days straight.
yes 
A little off topic:
We do it with NuPool. In our case apache2 acts as reverse proxy (port 80), which is passing requests from the clients to the hidden alp server (port 3333) and forwards the answers back to the client.
It’s a stability plus, because we can switch out the alp server any time if we need to, and it increases the security of our core infrastructure, because our web traffic is 100% behind CloudFlare CDN. DDOSing us is not impossible, but it’s much harder that way.
We’re also able to reroute our client traffic within 1-5 minutes to a backup server, if we needed too.
1 Like
I also use apache2, according to the guidelines specified by @woolly_sammoth. I do not use CloudFlare.
My LPs are having a hard time getting out of CNY, I feel. Remember that many of the alt coins on bter have cny pairs. I think someone could arbitrage BTS against poloniex right now with CNY back to NBT and make out on the deal, but I’d have to really do it out. With target reached on buy side, you could make more money getting back to the sell side anyway, so a small loss isn’t even that bad.
Alternatively, why are ya’ll selling on the btc pool and not the cny pool? If you don’t know about fill factor, we have a way (just ask if you don’t know 
) to semi-continuously adjust funds such that when someone sells you NBT for your BTC, it puts a portion of that over to sell for CNY. The price feeds are such that the LP would be selling BTC for CNY at 0.7% above the Bitfinex price, so on a micro scale it’s a good deal. From a macro perspective, splitting your funds between the pools will get you the most funds if both become fixed cost or regularly operate at target.
I am taking down the server for a little while. I’m moving to the bigger server for real now.
Edit: Up and running. I broke the website, but I’ll fix that later. I also wasn’t able to import the full wallet, so I just dumped any private keys that had any balances (I have plenty of network addresses for submitting liquidity).
Ugh, so many issues. Ok, so I got the site up, but the bots are all being funky and either not running or not crediting properly. Please stand by.
Got BTC up and running, I was just dumb and ran server.py instead of fc_server.py. I’m still having troubles with CNY, the pool is currently not running as no one can connect to it anyway. I suspect something to do with port 3333, but I’m not sure.
3 Likes
Still issues?
I still don’t have the CNY pool up. I will work on it again in a little while. I’ll need to switch it over to fixed cost once the grant passes anyway, but I will try to get it up and running with fixed reward in the meantime.
1 Like
No problem. Just saw that they were stuck. You do top notch work.
2015/11/11-15:52:57 INFO: bter - balance: 0.05963369 rate: 330.12% ppm: 4.56577188
AWESOME!
Thanks!
Things I’ve improved this time around (without the changes the grant will introduce):
- Bigger Server. Nud failed again last night, this should fix it.
- Fixed the deviation bug. This was serious, I’m really surprised we didn’t catch it earlier.
- Fillfactor now works on both buy and sell side. This constitutes a Tier 2 strategy.
- Shift is now adaptive. I’ll say more about what this means later.
- One more stab at the ordermatch bug, copying what sam did.
Most of these bugs are a combination of me having complicated ideals and not being a great programmer. I am trying not to spend a lot of time debugging old bugs like the order limit bug because of the parametric stuff coming out soon.
I assume you didn’t have a lot of funds in there? You shouldn’t be getting rates that high with any kind of volume (buy side is basically nonexistant right now).
@willy https://raw.nupool.net/nubits/ isn’t showing my BTC pool at all anymore. I think all I did was change IP’s, I’m still using the same custodial address.
2 Likes
Is your code on Github? I’d like to pull it to update my own server.
You mean your ALP server? Mine isn’t the best to use. You should use Sam’s for running server.py:
I use mine for the fc_server.py program though.
@nmei wait, how do you have a ppm of 4.5 but only a balance of 0.06? I thought ppm was basically NBT/minute. You shouldn’t be getting 4.5 NBT/min.
those very high values are due to a glitch. i see them also every now and then but i don’t trust them 
they don’t seem to have spiked his balance. Is the glitch only in the ppm or the rate % too? I don’t think my server is crediting improperly.
yes, it is only in those values. crediting is correct
I’ll look into it later. It seems like it’s an issue of transparency between the LP and the operator which, while important, is not critical. The fixed cost server has this issue in several ways at this point in time anyway. It’s hard to find out what the fc server is actually paying out right now without looking directly at the logs.
I got the CNY pool working on port 3335. I turned it off for now because I need to think about the best way to proceed / try to figure out what’s going on with port 3333.
nud set -maxconnections=8
OMG, my nud failed too…
Still not working on port 3335…