Closed Prossi79 closed 5 years ago
you shouldn't collect backtesting data from the same ip/server than running a live/dry-run bot ... they'll always impact each other.
The logs indicate no DDOs protection though - but as if the API was down / the network to the server was temporarily down ... (ddos protection rasises a DDoSProtection
error, not a RequestTimeout
).
@Prossi79, could you pls attach your config? (be careful. remove exchange and telegram keys and passwords)
@xmatthias .... mmmh, that is not very user-friendly :( I have just one Linux machine running, that is my VPS. I did not manage to run FT on my windows laptop.
Is there no way around this issue?
@hroff-1902 here is my config:
`{ "max_open_trades": 3, "stake_currency": "BTC", "stake_amount": 0.1, "fiat_display_currency": "USD",
"process_only_new_candles": true,
"dry_run": false,
"db_url": "sqlite:///tradesv3.prodrun.sqlite",
"unfilledtimeout": {
"buy": 10,
"sell": 30
},
"bid_strategy": {
"ask_last_balance": 0.75,
"use_order_book": false,
"order_book_top": 1,
"check_depth_of_market": {
"enabled": false,
"bids_to_ask_delta": 1
}
},
"ask_strategy":{
"use_order_book": false,
"order_book_min": 1,
"order_book_max": 10
},
"order_types": {
"buy": "limit",
"sell": "limit",
"stoploss": "market",
"stoploss_on_exchange": false,
"stoploss_on_exchange_interval": 60
},
"exchange": {
"name": "binance",
"key": "xxxx",
"secret": "xxxx",
"ccxt_config": {"enableRateLimit": true},
"ccxt_async_config": {
"enableRateLimit": false
},
"pair_whitelist": [
"ETH/BTC",
"XRP/BTC",
"LTC/BTC",
"BCH/BTC",
"EOS/BTC",
"BNB/BTC",
"XLM/BTC",
"ADA/BTC",
"TRX/BTC",
"BSV/BTC",
"XMR/BTC",
"DASH/BTC",
"IOTA/BTC",
"NEO/BTC",
"ETC/BTC",
"ONT/BTC",
"XEM/BTC",
"ZEC/BTC",
"VET/BTC",
"BAT/BTC",
"OMG/BTC",
"BTG/BTC",
"QTUM/BTC",
"WAVES/BTC",
"REP/BTC",
"DCR/BTC",
"LSK/BTC",
"RVN/BTC",
"NANO/BTC",
"ZIL/BTC",
"ZRX/BTC"
],
"pair_blacklist": [
"CRO/BTC",
"DOGE/BTC",
"XTC/BTC",
"MKR/BTC",
"USDT/BTC"
]
},
"experimental": {
"use_sell_signal": true,
"sell_profit_only": true,
"ignore_roi_if_buy_signal": false
},
"edge": {
"enabled": false,
"process_throttle_secs": 3600,
"calculate_since_number_of_days": 10,
"capital_available_percentage": 0.8,
"allowed_risk": 0.01,
"stoploss_range_min": -0.01,
"stoploss_range_max": -0.05,
"stoploss_range_step": -0.01,
"minimum_winrate": 0.30,
"minimum_expectancy": 0.10,
"min_trade_number": 10,
"max_trade_duration_minute": 1440,
"remove_pumps": false
},
"telegram": {
"enabled": true,
"token": "xxx",
"chat_id": "xxx"
},
"initial_state": "running",
"forcebuy_enable": false,
"internals": {
"process_throttle_secs": 4
}
}`
well - you risk bans - that's nothing to do with user-friendly or not - it's the exchange that's blocking you, not us. Add in a high enough limit ("enableRateLimit": true, "rateLimit": 1000
in the ccxt_async_config
section ) to avoid bans - but then downloading data will take forever since ccxt will pause 1 second between each subsequent call.
If you only download (nothing else running on that box) - then going to 200 (0.2s) will work for binance (i tested that downloading all of last years 1m data ...)
Also when you backtest, you eat away the machines resources, so a live/dry-run bot could suffer from that as well. Me personally, i would not risk that - however it's your decision, not mine.
For windows installation, you should try WSL, that's a linux subsystem (ubuntu by default) - so the linux installation guide does work. I think it's been in since windows 8 in experimental mode - but with windows10 you can install it from the windows store...
Thank you @xmatthias ! : ) it was not meant in a destructive way, but okay I understand the root cause issue now
didn't understand it that way - there's just nothing i would know off that we can do against that :disappointed:
I'll close this as this can be prevented using rateLimit, which is now also included in the sample settings.
Step 1: Have you search for this issue before posting it?
Step 2: Describe your environment
Step 3: Describe the problem:
I have tried to fetch data for backtesting on Binance ... 1m, probably too much
My bot on Binance running in production mode "died" short after, the one in dry run not
Steps to reproduce:
Observed Results:
I think the dead process was caused by the DDoSP of Binance ... but the bot should not stop, if I overload Binance by my requests, right?
Relevant code exceptions or logs: