[JBoss JIRA] Created: (TEIID-897) Create a new autowrap mode to replace pessimistic/optimistic
by Steven Hawkins (JIRA)
Create a new autowrap mode to replace pessimistic/optimistic
------------------------------------------------------------
Key: TEIID-897
URL: https://jira.jboss.org/jira/browse/TEIID-897
Project: Teiid
Issue Type: Feature Request
Components: Query Engine
Affects Versions: 7.0
Reporter: Steven Hawkins
Assignee: Steven Hawkins
Fix For: 7.0
There are issues with pessimistic/optimistic autowrap modes:
1. optimistic transaction autowrap is no longer needed since all installations of Teiid have a transaction manager available. This mode is really only useful as a development time aid - there is also a case to replace it with a readonly mode to prevent all updates.
2. the detection of when a transaction is needed is based upon the resolved from of a command and not the actual plan/execution. This has led to progressively more pessimistic logic to capture when a transaction is needed (being fully pessimistic we'd have to assume that any procedure with an update count of 1 also needs a transaction since an error can occur during the conversion of output parameters that should cause a rollback). We should either do a better job based upon the plan/execution, or just state simple assumptions (select -no txn, source insert, update, delete, exec w/ update count <=1 -no txn, else txn).
3. the terminology is confusing since it's evocative of locking semantics
I therefore propose we change these to txnAutoWrap=AUTO, which would be similar to pessimistic mode and would change the detection strategy to one of the proposals from 2.
We should also be explicit that this mode is based upon the assumption that multiple selects do not need to be grouped in the same transaction. However when there is a transaction, then the selects will be performed in the same transaction.
--
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators: https://jira.jboss.org/jira/secure/Administrators.jspa
-
For more information on JIRA, see: http://www.atlassian.com/software/jira
14 years, 6 months
[JBoss JIRA] Created: (TEIID-917) xml stream handling is inconsistent
by Steven Hawkins (JIRA)
xml stream handling is inconsistent
-----------------------------------
Key: TEIID-917
URL: https://jira.jboss.org/jira/browse/TEIID-917
Project: Teiid
Issue Type: Bug
Components: Common
Affects Versions: 6.0.0
Reporter: Steven Hawkins
Assignee: Steven Hawkins
Fix For: 7.0
Our SQLXMLImpl allows for a variety of source types - such as a streamsource coming from the xml source connector. However if the resulting SQLXML is used by the client (and not immediately consumed by something like XQuery) then the stream can only be read once. Also all of the conversion routines in the XMLTranslator logic (and in the xml document stylesheet application) are based upon full materialization. This may lead to memory problems. In general our SQLXMLImpl should be backed by a string if small and a (temp) file if large. Conversions from DOM and SAX sources would also need to create temp files to protect memory and allow for multiple reads respectively.
--
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators: https://jira.jboss.org/jira/secure/Administrators.jspa
-
For more information on JIRA, see: http://www.atlassian.com/software/jira
14 years, 6 months
[JBoss JIRA] Created: (TEIID-936) Query never returns from processing
by Paul Nittel (JIRA)
Query never returns from processing
-----------------------------------
Key: TEIID-936
URL: https://jira.jboss.org/jira/browse/TEIID-936
Project: Teiid
Issue Type: Bug
Components: Query Engine
Affects Versions: 7.0
Environment: fedora 12, Teiid build 1/19/2010, OpenJDK Server VM (build 14.0-b16, mixed mode)
Reporter: Paul Nittel
Assignee: Steven Hawkins
I'm executing TPCR Query #16. It processes correctly, and very quickly, in Designer, which is using an older Teiid query engine. The current Teiid, running via AdminShell, never returns from this query. Both are using the same VDB.
Two tables are involved PART (200,000 rows) and PARTSUPP (800,000 rows). The output will contain 18,314 rows.
The transformation query is:
SELECT
P_BRAND, P_TYPE, P_SIZE, COUNT(DISTINCT PS_SUPPKEY) AS supplier_cnt
FROM
TPCR.PARTSUPP, TPCR.PART
WHERE
(P_PARTKEY = PS_PARTKEY) AND (P_BRAND <> 'Brand#45') AND (P_TYPE NOT LIKE 'MEDIUM POLISHED%') AND (P_SIZE IN (49, 14, 23, 45, 19, 3, 36, 9)) AND (PS_SUPPKEY NOT IN (SELECT S_SUPPKEY FROM TPCR.SUPPLIER WHERE S_COMMENT LIKE '%Customer%Complaints%'))
GROUP BY P_BRAND, P_TYPE, P_SIZE
The query being executed is: select * from vTpcr.Q16 ORDER BY supplier_cnt desc, p_brand, p_type, p_size
I have tried to execute shorter version of the transformation query in AdminShell to no avail. This query: SELECT P_BRAND, P_TYPE, P_SIZE FROM TPCR.PARTSUPP, TPCR.PART WHERE (P_PARTKEY = PS_PARTKEY) AND (P_BRAND <> 'Brand#45') AND (P_TYPE NOT LIKE 'MEDIUM POLISHED%'); did not return in a reasonable time.
Teiid seems to use a fair percentage of the CPU for a while, then usage drops to ~0.1%
--
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators: https://jira.jboss.org/jira/secure/Administrators.jspa
-
For more information on JIRA, see: http://www.atlassian.com/software/jira
14 years, 6 months