SQL Anywhere 11.0.1.2376 Oracle 10g we have a sync environment running successfully... except for one table: - Inserts appear to always work fine - Updates randomly shift contents by 1 or 2 columns - appears to only affect CLOB columns - appears to only occur after preceeding NULL column content - we have compared our Publication, Insert and Update scripts (retrieved from ml_scripts table on the Oracle database) and believe they match as expected. (and... the shifting does not occur consistently) - we have verified we are using "iAnywhere Solutions Oracle - 11" ODBC driver on the MobiLink server connection to Oracle - this is a IIS Relay Server setup - the MlServer log file (running with -v cfhnpstur) shows the data content in the correct columns Not sure what else to look for?? Thanks, Bill asked 26 Jan '10, 19:22 Bill Aumen |
after Tech Support spent quite a bit of time working on this for us, it turned out to be unexpected behaviour (double speak for a "Bug") in our own AFTER UPDATE trigger. answered 09 Feb '10, 04:19 Bill Aumen 1
OK... share! (bugs in triggers are some of the hardest to find, and the engineer in me wants to know the details) Will do... we disabled the trigger for the time being as Lia is off for a few days and the trigger functionality is not really needed yet. |
Offhand, I'd say this needs hand-on and/or real-time study... perhaps it's time to grab a cuppa and call tech support. One thing you might try in the interim, is to determine if you can repeat the symptom using ordinary INSERT statements via SQL*PLUS, or not, thus proving it is definitely the fault of Oracle or likely the fault of the synchronization process (note the words definitely and likely... if the INSERTs work it could still be the fault of Oracle).
FWIW it's really really really really really unlikely to be the fault of Oracle, so don't read past the "call tech support" part of my previous comment :)
The Inserts work correctly, only the updates are screwy. After we just poured over our setup one more time, we have just opened a Support Incident. Will post the solution when we get it figured out. Thanks Breck.