Johan Hallgren commented on AMQ-660:
I've done some further testing:
+ First, with the broker stopped, I dropped the activemq_msgs and active_acks tables, and then recreated them manually using the SQL statements you provided. The delete statement that caused the error message then executes normally.
+ I then started the broker, and saw that the error message was still in the output. Manually running the delete statement also gave the same error, even though both tables still contains no data. I have attached a log dump for this startup (attached it anonymously by accident; didn't see that my session had timed out).
+ Finally, I studied the log dump and saw that there are three indexes being created on activemq_msgs on broker startup. Dropping the activemq_msgs_eidx makes the error message go away.
+ My initial thought was that the indices exceed MaxDB's length limit for table columns belonging to an index (1024 bytes), but that does not seem to be the case, and in any event it is specifically the activemq_msgs_eidx index that causes the error; dropping the other two indexes does not resolve anything.
So, in summary, I'll ask the MaxDB mailing list about the issue. In the meantime, do let me know if you have any other ideas that you want me to test.