we are running Attunity 6.4.0.425 to replicate (and transform) 109 tables from Oracle 11G to Redshift. For the last 6 weeks every couple days we get log messages that say processing has gone into one-by-one mode and then later an error message about unable to update a certain record. when we look in Oracle we see the record but it doesn't exist in Redshift - we are wondering if it is just not getting written to S3 or if we have some race condition due to the 109 tables we are replicating being split into 5 separate tasks so the update is happening before the insert. We have tried to fill in the missing records but ultimately we have had to reload the entire table into Redshift when we get missing records. We have been unable to reproduce this issue in our test environment and when we turn on verbose logging for any period of time we can't catch the problem before performance degrades and/or we run out of disk space due to all the logging.
↧