[Passed] Liquidbits.net trustless liquidity provision on USD, EUR and BTC pairs

[Cybnate]

Confirmed in the SSH logs that there was a memory problem, creating login issues and making the server non-responsive. Memory usage was at 93% when I managed to get through, normally at around 75%. Will keep an eye on this. There might have been a relation with the fork Nu had in the past or there is a memory leak. Will update NU with the next reboot. The new payout time has now shifted to around 11am UTC in case you wonder.

Apologies for the inconvenience.

[Cybnate]

I am not asking you to change your tolerance, I think %0.75 is fine. I simply am asking that you make your intended participants’ offsets to be relevant to that number. Please increase your Repo’s offset to 0.4% or call a spade a spade and reduce your tolerance to 0.5% so you can be fair about your 0.2% offset. Running with a 0.75% tolerance and defaulting the providers to a 0.2% offset is unfair to your providers who trust you to give them the best software for the job.

I’m not feeling comfortable to change code when I do not completely understand the implications as the change you propose is not in the config files and need to be made in the code.
Hope @woolly_sammoth can provide some advise on how Nu-pool deals or intends to deal with this.

[cryptog]

with on average 10% per month (0.33% per day) per month on Bitcoin.co.id BTC/NBT pair to a maximum of 495 NBT in total but split as follows:

Shouldn’t 495 NBT be in fact 5000 NBT?

Apart from that, since I am not clear 100% on the difference between spread, tolerance and offsets and since Liquidbits.net has produced a nice job so far, I will vote for its continuation.

I am waiting for the new hash.

[Cybnate]

Shouldn’t 495 NBT be in fact 5000 NBT?

Yes, it should. Well spotted. I’ve updated the draft.

[mhps]

I am trying liquidbits on a ubuntu machine. I get this error message upon starting
2015/07/26-xx:xx:xx ERROR: unable to receive balance for usd on ccedk: exception caught: 'bool' object has no attribute '__getitem__'

Any idea what is wrong? @woolly_sammoth @willy

[woolly_sammoth]

Not sure off hand but will look into it later today.

[mhps]

This might be relevant -
I tried with the same setup on a R-pi with no error.
I tried nupond with cny on the ubuntu machine and there was no error.

[mhps]

I might have found the reason – I was using a deleted (presumablly invalid) api and private keys on ccedk.

[mhps]

I see there are sell orders at higher price than 1.002 and buy orders at lower than 0.998. How do I set this spread and do I get bumped off earlier if the liquidity gets fully subscribed?

[masterOfDisaster]

This spread is one of the configuration parameters the tllp bot uses.
From @Nagalim’s NBT/CNY pool settings

less unix/nupond/pool-bter-cny
[...]
# Determines market spread
offset = 0.004
[...]

If you increase this from 0.004 to, say 0.006 the bot’ll place sell orders at 1.003 and buy orders at 0.997.
It depends on the parameters of the pool whether you’ll get rewarded for liquidity with that big spread.

[mhps]

Thanks. I didn’t know this was possible with LiquidBits bots. Looks like it is also allowed on policy level as https://raw.nupool.net/ counts these liquidity in.

[masterOfDisaster]

I’ve switched all my tllp bots to the version provided by @Nagalim.
I love this parameter in the config files:

# Sets time (in hours) between bot resets. [bug fixer]
restime = 1.00

It doesn’t always help to fix bugs where not all is put into walls (in times where the compensation limit is not reached!) as a result of changing prices.

Using the same bot software everywhere makes maintaining and troubleshooting much easier

…just to be clear: obviously each bot runs a dedicated copy of @Nagalim’s pool software!

[Cybnate]

Shameless plug, please vote for this: [Passed] LiquidBits term 3 Trusted Network Liquidity Pool

[Cybnate]

LiquidBits allows a 0.0075 offset to decrease order replacing due to rates swings. I’m aware that this wouldn’t be required for fiat pairs. However I can only have a global setting for all pairs on the server. For term 4 I’m thinking of setting up another server so I can have different server settings for the fiat pairs vs the BTC pairs. An alternative is to reduce the offset from 0.0075 to 0.0025 or so.

[Cybnate]

I’ve switched all my tllp bots to the version provided by @Nagalim.

Thanks for testing I’ve been holding off to ensure the changes made are stable and working.

[mhps]

I’m aware that this wouldn’t be required for fiat pairs.

Fiats can fluctuate by a few percent in a day, too. 0.002 on each side for EUR is alright. CNY floats within a band of few percent against USD.

[Cybnate]

Fiats can fluctuate by a few percent in a day, too. 0.002 on each side for EUR is alright. CNY floats within a band of few percent against USD.

I’m creating confusion again with this. The normal band stays 0.002 on either side. The 0.0075 percentage is what is allowed to deviate from that when rates change before the server shifts the order. I understand that with Nagalim’s version you are using that margin to place your orders wider than the standard 0.002 on either side. Apparently the server allows that.

[Nagalim]

An alternative is to reduce the offset from 0.0075 to 0.0025 or so.

Please don’t do that. I did that for a month, it gets buggy and weird. My CNY pool (I run 2 separate instances of server.py with separate custodial grant addresses, which is complicated and annoying) has a 0.6% tolerance: 0.4% offset, 0.19% deviation. This has been working wonders.

Your 0.75% tolerance is fine as a compromise between btc and fiat.

[seki]

when i try to run the bot i get the error:
"could not retrieve ccedk ids,server is unreachable"
i haven’t changed anything in the conf file, i only set my nbt address and API keys.
I already run a bot in cryptsy from the same windows rig.

edit: now it is working. it could be a temporary exchange issue!

[masterOfDisaster]

You might experience that ccedk is not as reliable as other exchanges in terms of API access
But good enough to let a bot run there