Hi Arcady,
W. 2014-11-10 15:34:11. <179> [10039] Error detected during multi-row operation, performing rollback before retrying in single row mode
The original warning (and yes, it's a warning and not an error in this case) is specifically because the upload operation that we tried to do as a "wide" operation (up to the number of rows specified by the mlsrv16 -s switch), failed. Because we recognize that the wide insert may be failing due to a *specific* row in the wide operation, we instead try each row individually to see if this resolves the problem (or provides more specific error details for which row is actually causing the problem).
Hence why this initial action is a warning - we're not sure if we can recover or not, and we're noting what we're trying to do to resolve the problem. If there is a subsequent error message after retrying, this is the actual problem (and we'll call handle_error/report_error, etc.).
W. 2014-11-10 15:34:11. <179> [10040] 111 row(s) were ignored in uploading data into table G4_cached_search_response_multi_prices
So, I need a clarification on this - was there an actual problem, or wasn't there?
This last warning is indicating again that there could be a problem, but you have specified upload_* scripts for this table that ignores uploaded rows from the remote. Did you really mean to send the rows from the remote, or did you really forget to implement the upload scripts? We don't know, so we just warn you that this is happening in your environment for you to examine later.
From a "synchronization" perspective, neither of these warnings would prevent the "Synchronization complete" message that should be logged after every successful synchronization. If the synchronization didn't complete, it should say "Synchronization failed."
---
What are your upload_* scripts for this table? Do you really want remotes to send up rows for these tables, or should the remotes have a "download only" action for these tables?
Regards,
Jeff Albion
SAP Active Global Support