Applies to:Product Release: V7 (Enterprise)
Version: 220.127.116.11  Fin Svcs
Database: IBM DB2 for zOS and OS/390 v8
Application Server OS: Microsoft Windows 2000 Server SP 4
Database Server OS: IBM zOS
This document was previously published as Siebel SR 38-3161028321.
In last maintenance release of Siebel (18.104.22.168) there is a feature introduced
related to canceling query that is taking long time once submitted by the user. The Cancel Query
feature lets the user end a lengthy query and continue working in the same Siebel session, while
also terminating the query on the database and application server.
We made necessary CFG
change and are able to get a pop up to cancel the query. Once the cancel button is invoked, the
pop changes into canceling query (see attached document). However, we are monitoring from OM logs
and DB that none of the resources are being released and IO on the database is still clocking. We
are using DB2 v8 without connection pooling in this environment, however in production connection
pooling is enabled. The attached are logs from OM as well.
Customer was testing the cancel query feature on Siebel SIA version 22.214.171.124 connecting through IBM DB2 Connect Enterprise Edition version 8 FixPak 10 to IBM Db2 for z/OS v8.
The behavior was reported as follows:
a) Start a long running query.
b) The 'running/ cancel query' pop-up box is shown.
c) I can see the running thread in Db2 using DB2 PM.
d) I cancel the query selecting the 'cancel query' option, the pop-up box changes to cancelling query'.
e) The original thread still runs in Db2.
f) The 'SBL-DBC-00101' error never comes back until the query finishes by itself, in other words, if the query runs for 2 hours by itself, the error comes back in 2 hours.
The user is unable to do anything else in Siebel until the 'SBL-DBC-00101' error comes up.
After extensive investigation by both Siebel and IBM Support it could be determined that:
1) There is an issue using the cancel query feature with Siebel connection pooling pointing to IBM DB2 for z/OS v8 RDBMS servers. Change request #12-TNT0Z0 was opened for this topic some time ago and the fix needed in order to leverage this functionality has been introduced in Siebel Maintenance Release version 126.96.36.199.
2) The behavior originally reported by the customer (without using Siebel connection pooling) could be identified as being due to the SQLCancel call taking place on a hostname/port that was not accepting requests. Therefore the SQLCancel was actually never issued on the RDBMS host.
IBM has a document discussing this type of occurrances:
Once this was resolved the cancel query feature was working as expected.