|
Snapshot Level Application |
Logical Data Grouping subsection |
Monitor Switch Statement |
|
Resettable |
No |
|
|
Element Name Element Type |
tq_cur_send_spills gauge |
Description: Current number of tablequeue buffers residing in a temporary table.
Usage: An agent writing to a tablequeue may be sending rows to several readers. The writing agent will overflow buffers to a temporary table when the agent that it is currently sending rows to is not accepting rows and another agent requires rows in order to proceed. Overflowing to temporary table allows both the writer and the other readers to continue processing.
Rows that have been overflowed will be sent to the reading agent when it is ready to accept more rows.
If this number is high, and queries fail with sqlcode -968, and there are messages in db2diad.log indicating that your ran out of temporary space in the TEMP table space, then tablequeue overflows may be the cause. This could indicate a problem on another node (such as locking). You would investigate by taking snapshots on all the partitions for this query.
There are also cases, perhaps because of the way data is partitioned, where many buffers need to be overflowed for the query. In these cases you will need to add more disk to the temporary table space.