Overview
Margin FIX uses standard FIX error messages with additional detail in the Text field. Errors fall into two categories:- Session-level errors: Protocol violations, handled with Reject (35=3)
- Business-level errors: Application logic issues, handled with BusinessMessageReject (35=j) or order-specific rejection messages
Error Message Types
Reject (35=3)
Used for session-level protocol violations.Session Reject Reasons (373)
BusinessMessageReject (35=j)
Used for application-level business logic errors.Business Reject Reasons (380)
Order-Specific Rejections
Order Reject Reasons (103)
In ExecutionReport (35=8) withExecType=Rejected:
Cancel Reject Reasons (102)
In OrderCancelReject (35=9):Common Error Scenarios
Example: Invalid TagOrder-entry failures returned by the exchange are sent as ExecutionReport (35=8) with ExecType=Rejected, not as BusinessMessageReject. BusinessMessageReject (35=j) is used for application-layer failures before normal exchange rejection handling, such as rate limiting or listener-session restrictions.
Troubleshooting
MsgSeqNum Too High on Logon
Symptom: Logon fails or the server sends a ResendRequest for messages the client doesn’t have. Cause: The client is sending aMsgSeqNum higher than what the server last saw. This typically happens when the client’s local sequence store persists across sessions but the server has reset (e.g. after maintenance or a prior ResetSeqNumFlag=Y logon).
Fix:
- KalshiNR, KalshiDC: Set
ResetSeqNumFlag<141>=Yon every Logon. These sessions require it; Logon will be rejected without it. - KalshiRT: If you don’t need to recover missed messages, set
ResetSeqNumFlag<141>=Yto reset both sides to 1. If you do need retransmission continuity, ensure your local sequence store matches the server’s state.
ResetOnLogon=Y in your session config for non-retransmission sessions.
SendingTime Rejected
Symptom: Reject (35=3) withSessionRejectReason<373>=10.
Cause: The client’s clock is more than 30 seconds off from the server. Sync your system clock via NTP.
Duplicate Session (“already exists”)
Symptom: Logout (35=5) immediately after Logon withText<58>="already exists".
Cause: Another FIX connection is already active with the same API key and TargetCompID. Only one connection is allowed per API key per session type. This can also occur if a previous connection was not cleanly closed and the server hasn’t yet detected the disconnect.
Fix: Ensure the previous session is fully disconnected before reconnecting. If the prior connection was lost unexpectedly, wait for the server’s heartbeat timeout to expire (up to 60 seconds depending on your HeartbeatInt setting) before retrying. Use separate API keys for concurrent connections.
Logon Signature Rejected
Symptom: Logout immediately after Logon with a signature error. Cause: TheSendingTime used in the pre-hash string doesn’t match the SendingTime<52> in the actual Logon message. If using a FIX library, the library may auto-populate SendingTime. Use that exact value when computing the signature, not a separately generated timestamp.