Skip to main content
Question

Issue with Unidata EDA / SQL error.

  • August 24, 2026
  • 2 replies
  • 70 views

Darren Johnson
Forum|alt.badge.img+1

Getting the below issue operating replication with EDA.

Hoping someone can shed some light on how to resolve our issue below.


U2\BIN\RW.ERRLOG u2\bin\rw.errlog

EDA SQL Server Driver: [Microsoft][ODBC Driver 18 for SQL Server]TCP Provider: An existing connection was forcibly closed by the remote host.

Mon Aug 24 06:20:53 
RW(0,7460) report:
Mon Aug 24 06:20:53 
RW(0,7460) report:
In E:\U2\sys\CTLG\c\CXG.WRITE.EDA.BI.DATA at line 404 Mon Aug 24 06:20:53 
RW(0,7460) report:
EDA_write_tuple error, id = 'A143088'
Mon Aug 24 06:20:53 
RW(0,7460) report:


The EDA file is CXSFILE stored on the Unidata Subsciber. We get the above error attempting to write to CXSFILE (EDA file) with a
Trigger on CXFILE as below configuration.

CREATE.TRIGGER CXFILE CXG.WRITE.EDA.BI.DATA AFTER UPDATE
CREATE.TRIGGER CXFILE CXG.WRITE.EDA.BI.DATA AFTER DELETE

Some evidence supports: client-side dead-handle defect, stale held connections are not detected and never re-established.

I believe we have covered off root cause with Subscriber (Server configuration, Network configuration, Collexus configuration, EDA configuration, Virus scanning configuration) & SQL (Server configuration, Dead locks, Table configuration).

The below outlines the sequence of events.

Trigger fires -> EDA_write_tuple -> Several successful writes -> EDA_write_tuple fails -> Trigger process is now in error state - > Subsequent trigger write fail.

Normal Unidata write -> Successful.

UniData 8.3.1
     ↓
EDA 4.33.0
     ↓
Microsoft ODBC Driver 18 for SQL Server
     ↓
SQL Server 2022.
 

2 replies

John Jenkins
Forum|alt.badge.img+1
  • Participating Frequently
  • August 24, 2026

Darren,

 

Use SQL Server Management Studio under Management→ SQL Server Logs​​ I sospect you will see some log errors around the time of the event​ ​​. I.ve seen multiple causes of SQL erver dropping a connection mid-update and these include:

 

  • User permissions on the physical files
  • SQL permissions on the table and linked tables (associated multivalues are linked tables in SQL Server) for  the user ID used in the cinnection properties.
  • SQL Server cache memory size exceeded (A SQL Server tuneable).

or on the U2 database side - causing an error in SQL Server:

  • Invalid data for the data type and restrictions defined for the SQL table
  • Incorrect of missing associations for multi-values and multi-subvalues 

I am assuming that you have already tried a test migration of data with success and thet the version of the MV database you are using is certified against SQL Server 2022? If  not the I’d start here with a  controlled test migration.

Regards

JJ


Darren Johnson
Forum|alt.badge.img+1
  • Author
  • Participating Frequently
  • August 30, 2026

Hi John,

 

I’ve reviewed your response to the question, but I’m still unclear on why the trigger continues to fail.

What I’m trying to understand is why the process executed from ECL is successful after the tuple error, whereas the trigger continues to fail under the same circumstances.

Can you please clarify what specifically causes the trigger to fail and why it is unable to recover after the tuple error?

Thanks for your support