Combo bot grid stops cycling silently — lastPrice drops to 0, UI shows "connected"

Hello Gainium Community,

I think I might be affected by a bug, but I’m not that experienced with combo bots so I wanted to check here first in case I’m missing something.

I have an ETH/USDT combo bot on Binance Spot (V1 interface). It’s 4 days old and the price feed seems to silently die — lastPrice drops to 0 via the API, the combo grid stops cycling, but the UI still shows “Websocket connected.” No errors shown anywhere.

What I’m seeing:

  • Bot opens a deal, places combo grid orders — all normal initially

  • At some point the grid stops cycling. Sells fire but no corresponding re-buys are placed

  • Binance open orders shows only DCA buy orders far below price — zero sell orders near current price

  • The API shows lastPrice: { "ETHUSDT": 0 } and the updated timestamp goes stale (9+ hours without updating)

  • The only fix is a manual stop/restart

This has happened on both deals the bot has run since I created it. Each time, the grid was dead for hours before I noticed. On restart, all the pending sells fire as a burst, the trailing TP triggers, and the deal closes — but the grid profit from the dead period is lost.

I also noticed an orphaned notEnoughBalance flag on a sell order belonging to an already-closed deal, which seems like it shouldn’t be there. Not sure if that’s related.

Is this a known issue, or am I likely configuring something wrong? Happy to share my full config, API responses, and Binance order history if that helps narrow it down.

Thanks in advance!


Detailed bug report

Bot details

  • Bot type: Combo bot

  • Pair: ETHUSDT

  • Exchange: Binance Spot

  • Interface: app.gainium.io (V1)

  • Bot ID: 69ced26f1004d803c3c7b8eb

What happens

  1. Bot opens a new deal, places combo grid orders (buys + sells) — all normal

  2. Initial grid orders fill correctly on Binance (confirmed via Binance order history and Koinly)

  3. At some point after the initial orders fill, the price feed dies silently

  4. lastPrice drops to 0 (confirmed via API)

  5. The bot stops cycling grid orders — no new buys or sells placed despite price crossing grid levels repeatedly

  6. The UI still shows “Websocket connected” — no indication anything is wrong

  7. The bot’s updated timestamp goes stale (9+ hours without updating)

Evidence

Via API (GET /bot):


"lastPrice": { "ETHUSDT": 0 },

"lastUsdRate": { "ETHUSDT": 0 },

"usdRate": { "ETHUSDT": 0 },

"updated": "2026-04-06T00:00:18.891Z" // 9 hours stale at time of discovery

On Binance:

  • ETH was trading at ~$2,131 while Gainium showed lastPrice: 0

  • All open orders on Binance are BUY orders only (DCA grid) — zero sell orders near current price

  • Binance order history confirms the initial grid sells filled correctly at 01:00 UTC

  • No replacement sell orders were placed after the initial fills

Possible related issue — orphaned order on closed deal:


"notEnoughBalance": {

"orders": { "69ced273d3fa3b08d5de6dd5@SELL@211054": 1 }

}

Deal 69ced273 is already closed, but the bot is still tracking a sell order for it with a “not enough balance” flag. This stale state from a closed deal might be what’s crashing the price feed.

Occurrence pattern

Bot is 4 days old. Feed has died on both deals it has run.

  • Incident 1 (Apr 3–6): Feed died shortly after bot was first started. lastPrice showed 0 for ~3 days. Initially thought Binance was executing orders fine (Koinly showed fills) — but those fills were from the initial order burst, not ongoing grid cycling. Grid was genuinely not cycling. Recovered by restarting the bot. Deal closed at +$0.55.

  • Incident 2 (Apr 6): New deal opened at 01:00 UTC after incident 1’s restart. Feed died again within minutes of the new deal opening. Same symptoms — lastPrice: 0, no grid cycling, UI shows “connected.” Ran for 9 hours undetected. On restart, the chart showed all pending sells firing as a burst but zero corresponding buy re-entries — confirming the grid was not cycling (sold without re-buying). Deal closed at +$0.199 (should have been much more with 9 hours of grid cycling at those levels).

Both times, the only recovery was a manual stop/restart.

Other observations

  • An orphaned notEnoughBalance flag was found on a closed deal’s sell order (see above) — possible trigger for the feed death.

Config

Standard combo bot config — no custom indicators, no webhooks, no unusual settings. Long strategy, 5 base grid levels at 1.5% range, 8 DCA orders, trailing TP at 0.5% deviation. Happy to share full config if needed.

Expected behaviour

  • Combo grid orders should cycle continuously (buy and sell at each grid level crossing)

  • If the websocket drops, the UI should surface an error rather than showing “connected”

  • Closed deals should not have orphaned orders tracked against the active bot state

1 Like

We have received your bug report regarding the Combo bot grid stopping cycling silently and other related issues. We appreciate you providing such detailed information, including the bot ID 69ced26f1004d803c3c7b8eb. We are looking into this and will provide an update shortly.

1 Like

Your binance API keys are outdated. They should be in a new format (ed25519)

You dont need to create new exchange, simply update keys.

With these keys we are not able to receive order updates

1 Like

Confirmed all good, thankyou Maksym :slight_smile:

Hey Maksym, team, this does seams to be re-occuring for me afterall.

Update — 3rd incident today (8 Apr)

just wanted to update — this has happened again, 3rd time in 6 days. Same pattern: grid was cycling fine, then at some point around 07:24 UTC it stopped silently. ETH was ranging between 2240–2255 crossing active grid levels for 5+ hours with no fills. Manual restart fixed it again immediately.

I’ve been pulling the API data each time this happens. Here’s the before/after from today’s restart:

Dead state (07:24 UTC) After restart (12:18 UTC)
updated timestamp 5+ hours stale Fresh
ETH held 0.0139 (static) 0.0162 (moving)
USDT held 505.56 (static) 500.42 (grid cycling)
lastPrice 0 0 (seems cosmetic — always 0 for me)

I also noticed an orphaned notEnoughBalance entry that references a sell order on a deal that’s been closed for days:

"notEnoughBalance": {
  "orders": { "69ced273d3fa3b08d5de6dd5@SELL@211054": 1 }
}

This has survived every restart across all three incidents. I don’t know if that’s normal or something that could be related, but it’s the one constant.

I’m still fairly new to combo bots so I could easily be misconfiguring something. Are there any settings, logs, or diagnostics I should be checking on my side to help narrow this down? Happy to share my full config or run any tests if it helps.