Iโve built a custom live market dashboard using Fyers WebSocket (SymbolUpdate
+ DepthUpdate
) for full F&O symbol coverage (~220+ stocks). The system tracks and logs tick-level trade and depth data continuously during market hours, with resubscription logic in place for reconnects and symbol silence monitoring.
โ๏ธObserved Issue
After running the system at 12:02 PM, I noticed that only 180 out of 251 subscribed symbols were receiving trade data, despite:
โ Full WebSocket connection success (
on_connect
triggered)โ Full symbol list resubscribed via
fy_socket.subscribe(...)
โ F&O symbols involved (which are highly active during the day)
โ 71 symbols received no trade updates post 12:02 PM, even after full restart
๐ DiagnosisTo confirm whether the problem was on my end:
โ I programmatically loaded the latest
trade_log_*.csv
.โ Filtered all rows after
12:02:00
.โ Counted unique symbols receiving data.
โ 71 out of 251 subscribed symbols were missing.
This proves that:
The client-side subscription was sent and logged.
No trade message was received for 71 active F&O symbols.
WebSocket remained stable โ no disconnects, no errors.
๐งช Technical Setup Summary
โ
fy_socket.subscribe(symbols, data_type="SymbolUpdate")
โ Reconnect logic and re-subscription chunks implemented
โ Symbol silence monitor runs every 10s with 30s timeout
โ No data ever arrived for ~70 symbols after successful subscribe
๐ก Hypothesis
This behavior points to a possible upstream issue in Fyers' WebSocket back-end, where:
Some subscribed symbols fail to deliver live ticks despite acknowledgment of subscription.
This occurs without triggering any error, making it hard to detect without a watchdog.
โ Recommendation / Request
Could the Fyers team:
๐ Review the WebSocket delivery guarantee for large symbol lists?
๐ฌ Ensure data delivery for all subscribed symbols, especially F&O?
๐ ๏ธ Consider emitting confirmation or error per symbol subscription (like a subscription acknowledgment)?
๐ Additional Info
Tools: Python, Dash, custom resubscription + monitor logic
Date of issue: 19/06/2025
Time of incident: Around 12:02 PM IST
Workspace stable (no network or CPU lag)