I would verify there is nothing changed in the network as well and maybe try uninstalling and reinstalling the drivers. I would completely remove them as well. The reinstall only one driver that is necessary for the Oracle Data source.
Hope this steps help!
Has this been resolved? We are running into the same error once upgrading to Oracle 12.2.
Hi Patrick - Apology for late response , We had to roll back the DB instance and couldn't test the suggested steps from you. We did install the latest driver after uninstalling the existing one, still no luck! Thanks!
Not yet.. Have you been able to fix it?
We have the same issue - upgraded from 11.2 to 12.2 - and get these every few seconds in the Oracle alert log. I opened an SR with Tableau Support a few months ago and was not able to get anywhere. I was advised to check Oracle Doc ID 2329165.1 and check the Tableau Oracle drivers but that did not resolve the issue. They are saying it's an Oracle issue, and Oracle says it's a Tableau issue. Not much help or progress, hoping your thread can help us though...
We are having the same issue on 184.108.40.206:
Errors in file [sanitized] (incident=10359):
ORA-03137: malformed TTC packet from client rejected:        
Incident details in: [sanitized]
Session (246,42034): Bad TTC Packet Detected: Inbound connection from client
Session (246,42034): Bad TTC Packet Detected: DB Logon User: [sanitized], Remote Machine: [sanitized], Program: tabprotosrv.exe, OS User: tableau_service
Session (246,42034): Bad TTC Packet Detected: Client IP Address: [sanitized]
Database upgraded to 220.127.116.11
The issue was investigated in BUG 26948049<https://support.oracle.com/epmos/faces/BugDisplay?parent=DOCUMENT&sourceId=2329165.1&id=26948049> - ORA-03137  FROM OCI APPLICATION, closed as not a bug. Issue occurs when setting 'maxarr_len' for non array input. In 12.2 were added array sanity checking. This is not a bug and is expected behavior.
It is found that the OCI program was setting 'maxarr_len' to 1 ('args.numRec') in the call to 'OCIBindByPos'.
Error is resolved after changing it to 0. Example:
status = OCIBindByPos(stmt,
i + 1,
0,//args.numRec, <<<<<<<<<<<<<<<< 'args.numRec' used to return 1 and was replaced with 0.
I don't know yet. We just upgraded our server to 2018.2 last night and we're testing it now.
I will let you know whether it works.
Thanks for the update!