RDBPROD: CODASYL DBMS DATA PAGE/SNAPSHOT PAGE DEADLOCK
Can anyone shed any light on this. We are seeing pages in data area 3 deadlocking with pages area 9 (the snapshot area for area 3).
This is is what was observed.Can anyone explain this?
It appears that of the two pages involved in the deadlock, there is always one in a data area and the other in a snapshot area.
Consider the following log extract, where the first timestamp in a header is the run time (coincidentally also every 3 seconds), and the second timestamp is the start of the lock. The deadlock seems to either “seesaw” or remain the same for several sets of 3 seconds. I have grouped the entries by run timestamp, and added a description of what they imply.