Test region looped 30 min on a DB2 cursor - how do you kill this fast?

Dev entered a bad key in test. CICS program opened a cursor, FETCH loop never set EOF, and it just sat there dumping to CEEMSG. Ops took forever to cancel the region because of the ticket process.

ICVR is 3000. There is an EXEC CICS PUT CONTAINER inside the loop so I am not sure runaway detection even helps.

Besides "fix the missing MOVE to WS-EOF" - what do shops do so one bad test does not pin a region? ASULIMIT? hard fetch caps in test only?

Fetch loop must key off SQLCODE, not a flag some other SQL might clobber. In test we also hard-cap fetches (e.g. stop at 200) when the region name is the test LPAR.

ASULIMIT from the DBA side can stop a runaway SQL. Ask them what it is set to in test vs prod.

Cancel the task from CEMT if you can still get in. Do not wait for a full region recycle unless the region is already wedged. We have a short-path for test cancels that skips the prod severity ladder.

Pushed for a test-only cancel path and a fetch cap. DBA is checking ASULIMIT tomorrow.