I encountered a strange situation regarding workload management. We are running a surrogate key generation stored procedure which does an ALTER TABLE to add the empty surrogate key columns at the begin of the process. The columns are added to the populated Stage Table. Usually this works quite good, but recently we loaded a table which very large rows. This leads to very ALTER TABLE runtimes. Furthermore, this statement consumes a lot of CPU and IOs.
In workload management we have a rule defined which aborts the session after it consumed more then 10.000 CPU seconds. As we hit this limit, this seems to happen (i can see the rules being triggered as well in the TDWM Qry Log Views). But as ALTER TABLE statements can't be aborted, it looks to me as if the workload management rule is applied but the ALTER TABLE is finished first and AFTERWARDS the session is aborted.