NuPond

[masterOfDisaster]

Just for the record:
I received payouts of 0.0301 NBT - so that is working as well!

Looking at the log I found:

ERROR: unable to retrieve order book for btc on bter: exception caught: [Errno -2] Name or service not known
WARNING: need to resynchronize trading bot for btc on bter because the deviation reached 29.69
INFO: waiting 6.48 seconds to synchronize with other trading bots for btc on bter

Does that point to a problem on my RaPi of is it the bot?

edit:
I checked old logs. This error is happening several times per hour. Hmm…

edit2:
The nupool client by @creon produces error as well, but only rarely and due to timeouts (“The read operation timed out”).
As this nupool client runs on a different RaPi it could still be RaPi related issues…
Don’t worry too much (unless I’m not the only one with these kinds of errors) and for now consider it “just for the record…”

[creon]

“Name or service not known” really sounds like your own internet was down for a while. Your Raspi 1 with several coin daemons could lead to this effect. I had the client running on a raspi 1, but without anything else running there, and I also had some more errors than on a PC (but still only maybe one in 30 min).

[Nagalim]

I’m pretty sure I understand every change I made to the code, and you shouldn’t really be getting any exotic errors. The worst that should happen is either the bot will be shifting the orders around too often or your rate won’t register. I’ve been running it all night and didn’t experience either of these problems, so I believe the update is stable.

If you are testing, you should set ‘prefunit’ to anything other than nbt or btc (leave it blank) so that your bot uses the widest spread. This will test the outer limits. I have it on ‘btc’ by default.

[masterOfDisaster]

you shouldn’t really be getting any exotic errors

Like I expected - the problem is on my end. Will move the client to a less utilized RaPi soon

If you are testing, you should set ‘prefunit’ to anything other than nbt or btc (leave it blank)

Just to make sure - are you speaking of that part of pool.conf

### GENERAL OPERATION PARAMETERS - DO NOT CHANGE ###
server=45.55.253.18:3333
unit=
exchange=bter

[Nagalim]

No, I’m talking about here:

OPTIONAL PARAMETERS

    # interest: minimal interest rate at which you want to get payouts in percent (0.25% = 0.25)
    # Set to 0 for pool maximum. You can specify two interest rates for the bid and ask side 
    # by separating them with a comma, e.g. 0.25,0.1
    interest = 0
    # ordermatch: place orders even if they would match with an existing order
    # set this value only to True if you know what you are doing.
    ordermatch = False
    # preferred unit uses a shorter spread on the unpreferred side. (nbt, btc, or blank)
    prefunit = btc

The last value is ‘prefunit’

[masterOfDisaster]

Got it. Will give it a try. Was asking because the “DO NOT CHANGE” in that part of pool.conf made me cautious
Maybe I’ve downloaded the wrong branch - my pool.conf doesn’t have

# preferred unit uses a shorter spread on the unpreferred side. (nbt, btc, or blank)
prefunit = btc

…
edit:
I’ve downloaded the branch “Nagalim-prefunit-1v” to the RaPi that is running successfully @creon’s nupool client for some time.

Settings for nupond pool.conf are:

# preferred unit uses a shorter spread on the unpreferred side. (nbt, btc, or blank)
prefunit =

Running smoothly, but I still get:

[Errno -2]

This is unique for the nupond client. Neither @cybnates liquidbits client nor creon’s nupool client ever produced that error so far.
I’ve grepped through all logs I have…

I’ve found nothing like “errno” in the liquidbits logs.
And the only numbered errors I found at the nupool logs are 8 (SSL related), 22 (debug message), 97 (address family not supported).
Only the nupond client creates “errno -2” and still in numbers…

Is it really related to my client or could it be something on bter’s side?

exchanges.py uses data.bter.com for the connection. Replacing that with 141.101.121.207 for a test run didn’t work:

ERROR: unable to delete all orders for btc on bter: exception caught: No JSON object could be decode

Is it because of cloudflare or do I have to adjust more?

[Nagalim]

Btw, your ‘unit’ in the general operations parameters section should definately be ‘btc’, if it isn’t already. The bit you posted has it blank.

[Nagalim]

Only the nupond client creates “errno -2”

I have not seen that. I’ll check it out later tonight when I get home from work.

Do you see the “errno -2” when you use innuitwallet on my server? I.e. is it just my experimental version of the nu-pool directory that throws this error, or is my server throwing this error despite what client you use?

[masterOfDisaster]

I have it with the older version as well. I had not investigated the logs before, because all was running fine - at least fine enough not to dig in logs
Here’s an example of the non experimental version of the nupond bot:

2015/05/01-11:19:31 ERROR: unable to retrieve order book for btc on bter: exception caught: [Errno -2] Name or service not known
2015/05/01-11:53:20 ERROR: unable to retrieve order book for btc on bter: exception caught: [Errno -2] Name or service not known
2015/05/01-14:17:37 ERROR: unable to retrieve order book for btc on bter: exception caught: [Errno -2] Name or service not known
2015/05/01-16:30:52 ERROR: unable to retrieve order book for btc on bter: exception caught: [Errno -2] Name or service not known
2015/05/01-16:43:14 ERROR: unable to retrieve order book for btc on bter: exception caught: [Errno -2] Name or service not known
2015/05/01-17:17:15 ERROR: unable to retrieve order book for btc on bter: exception caught: [Errno -2] Name or service not known
2015/05/03-21:43:03 ERROR: unable to retrieve order book for btc on bter: exception caught: [Errno -2] Name or service not known
2015/05/03-21:49:16 ERROR: unable to retrieve order book for btc on bter: exception caught: [Errno -2] Name or service not known
2015/05/04-05:17:15 ERROR: unable to retrieve order book for btc on bter: exception caught: [Errno -2] Name or service not known
2015/05/04-08:05:11 ERROR: unable to retrieve order book for btc on bter: exception caught: [Errno -2] Name or service not known
…

The server I use is:
server=45.55.253.18:3333

[Nagalim]

Ok, this issue is not with the experimental branch, it’s an issue between you and bter I think.

[creon]

This is a trading bot error, not a server error. Again, “Name or service not known” is what you get if you unplug your router and open a website. It really means “You don’t have internet, why are you asking me to open a connection?”.

I googled a bit and if this only happens this rarely then some people could improve their situation by adding additional DNS servers. But actually in your case it is really not that often. If you want I can disable the error message for you.

[masterOfDisaster]

Thank you for your offer, but I don’t think it’s necessary at all.
In the end the bot is running fine. Those few errors don’t hurt at all.
I was only trying to report findings that my help improving the bot…
…looks like I need to improve my infrastructure

It’s a bit strange though, that the same Raspberry that meanwhile hosts the (experimental) version of the nupond bot doesn’t have problems running the classic nupool bot.
As the DNS settings are the same for both bots I can’t explain why one bot throws these errors and the other doesn’t.

[Nagalim]

hosts the (experimental) version of the nupond bot doesn’t have problems running the classic nupool bot.

I thought you said it had problems with the classic nupond bot? If you run https://github.com/Nagalim/nu-pool on bter, do you get this issue?

[masterOfDisaster]

The log excerpt in one of my recent posts here in this thread is from the classic nupond bot running on bter.
It was running on a RaPi which also hosts some other applications.

I’ve recognized the errors after having started the experimental version of nupond bot.
Later I found out that the non-experimental version had created these errors as well.

I meanwhile moved the experimental version of nupond bot to a RaPi2 which hosts @creon’s nupool bot which has never reported such an error - the nupool bot is running ()more or less) continuously since 23rd April.

I still believe that it’s not worth looking into this any further as long as I’m the only one with that error (that doesn’t even hurt the operation of the bot)…

[Nagalim]

I restarted the server. It now pays out at 1 NBT 1/day.

I haven’t done the manual payments yet, but I have the credits files for all open balances. Your balance should reset to zero.

[masterOfDisaster]

I’ve started a test run with @creon’s bot using your server (45.55.253.18:3333) and bter (running at the same RaPi2 where @creon’s nupool but operates flawless on bittrex).

I expect to see the same error (“errno -2”), because I’m almost sure it has to do with bter (and maybe in some way my internet access…).
We’ll see.
I’ll keep you informed, but will be away most part of the weekend with no feasible access to my RaPis.
You might need to wait until anytime on Monday for a report.

[edit]
Got the first “errno -2” after less than 20 minutes.
So it has nothing to do with the nupond bot - it might have to do with bter or with my client. As the same machine is running fine with @creon’s nupool bot on bittrex, I wouldn’t wonder if other people got “errno -2” from time to time when running a bot on bter.
As this doesn’t seem to impact the bot operation, I’m going to ignore it
[/edit]

Thanks for your support and for providing nupond!

And thank you, @creon, for having created the nupool bot to make TLLP possible!

[Nagalim]

I fine-tuned trading.py to outperform other LPCs on bittrex for nupool:
https://github.com/Nagalim/nu-pool/tree/Champion

I also created on that’s more gentle for markets that haven’t reached target yet:

Nagalim/alp-collection

These branches are basically abusing the current config.py settings of TLLP servers, so limited time offer. They do not offer any new mechanics, just a tuning of the spread and perfunit parameters.

Edit: Ok, they are now updated and running properly, I think. Spread = .007, max deviation = .0015, prefunit = .0061 for champion and .002 for grass.

[Nagalim]

Alright, I manually distributed from before the grant passed. I’ll be restarting the server again this weekend to change the sampling. I will do a manual distribution for that after the restart.

Edit: Increased sampling to 24, restarted server, manually distributed.

[Nagalim]

The new server is:
NBT/CNY:
45.55.228.61:3333
NBT/BTC:
45.55.228.61:3334

I am not processing payouts, but the rates and targets reflect the beginning draft of nupond operation. This server will be linked from nupond.net after the grant passes.

[Nagalim]

I’m looking for testers for all pools. This update includes a modified directory structure and 3 new parameters:

1. offset (used to be spread)
2. deviation (keep between 0.001 and 0.003, make as large as possible)
3. restime (resets every 15 minutes to keep bugs away)

All three of these can be left out of the config file if you want (defaults: 0.002,0.0025,24 works on all operating pools). Note that all config files are now tuned to the pool.

I also fixed the executable and notepad issue for unix and windows directories respectively that @cybnate’s branch was having issues with.

Nagalim/alp-collection

I’m sorry to @woolly_sammoth and @willy for >70 commits. Me and github are currently in an abusive relationship. Hopefully we will resolve our issues soon, you don’t have to pull from my branch if you don’t want to. I made 0 server side changes, and I think all the client changes I made are ultimately beneficial.