Archives of the TeradataForum

Message Posted: Tue, 15 Feb 2011 @ 20:26:50 GMT

	<Prev	Next>		<<First	<Prev	Next>	Last>>

<-- Anonymously Posted: Tuesday, February 15, 2011 14:15 -->

Yesterday, a user executed a script that launched many single-query sessions sequentially, and in rapid succession (i.e. as soon as one session ended, it immediately launched the next session). During one time interval of approximately 30 minutes, 7 queries failed with a no more spool error. The DBQL data for these queries shows that they did NOT, individually, hit the user s spool limit. The DBQL data also shows that these queries did NOT overlap; that is, they executed sequentially; there was NO concurrent execution. So, why did these queries receive a no more spool error? I have two theories:

1. Spool is released asynchronously by a background task. For example: although Query_1 finished before Query_2 was launched, Query_1 s spool was still in the process of being released by a background task while Query_2 (and perhaps Query_3 and Query_4 ) was executing. The spool limit was reached because of the combined, concurrent spool usage of these queries.

2. When a query fails, DBQL does not record certain information, such as spool, accurately. So, although the DBQL data shows that the individual queries used far less spool than the user s spool limit, they actually used much more than that and they did, in fact, hit the spool limit.

Can someone tell me if either of these queries is correct?

	<Prev	Next>		<<First	<Prev	Next>	Last>>