If you can spare some time, please advise me on OTN as per the link:

http://forums.oracle.com/forums/thread.jspa?threadID=1095206&tstart=0

Thanks.

Can you please comment to my question on OTN as per the link below?

http://forums.oracle.com/forums/thread.jspa?threadID=1052993&tstart=45

Mr Hans Forbrich on OTN has answered my question and advised that you may have ideas based on your experience.

Thanks.

I was referring to one of your posts: http://www.freelists.org/post/oracle-l/RE-Calculating-LIOs,11. Yours was the only post on the entire internet that had some information on index prefetching. And I feel I am facing some issue related to the same. We are on 10gR2 10.2.0.4.0.

Did set _db_file_noncontig_mblock_read_count=1 but fetching did not get disabled. In our case, there are 2 queries – one doing an index unique scan and other doing an index range scan and I see “db file scattered reads” for both. Initially I was surprised to see scattered reads for unique/range scan of index but later understood (not sure if my understanding is 100% correct because you mentioned that “db file parallel read” should be seen) that it was index prefetching. And I want to disable that because it seems that on RHEL ext3 file system it is not ocurring but it is ocurring on solaris (sparc & x86_64 both) ufs and that is probably slowing down the system during the initial few minutes. Once it stops (probably by then the cache is warmed up) then the performance is fine.

Thanks

I’ve been developing for some time a tool named Mandela to diagnose performance problems. It need not be installed or create any object in the database analyzed. It only needs read access to the tables of STATSPACK or AWR.

The site is still under development, but if you want can download the Mandela and follow the launch of the latest versions.

What makes Mandela nice? Well, he can show into timeline all statistical system, redo log files switches, statistics I/O (tablespace, and datafile filesystem) and much more…

But what it does well done is to decompose the wait events (two clicks, three at most) and allow drilldown for the top sql by metrics (buffer gets, diskreads, cpu time, elapsed time, etc.) and now for the wait event. That’s right … from a chart of wait events you right click mouse and see the statements SQL that waited for “db file sequential read” or “CPU”, for example.

If you’re interested (and available) to see how Mandela can help you gain productivity, access http://www.trevis.com.br/projetos/mandela/download

If you want to chat or ask questions about the Mandela my contacts is as follows:

[] s,
Rafael Trevisan
http://www.trevis.com.br
EMail / MSN / GTalk: rafael@trevis.com.br

I was playing with the “*ss.awk” systemstate decoder script from LTOM, but frankly I would prefer to get first more knowledge about this kind of dumps. I’ve read in some section here that you had previously taught in a seminar ways to work with them, and I wonder if this particular seminar is still available online?

Best regards, and great thanks for your site and effort!

Have emailed you @ tanel@tanelpoder.com

Cheers,

Simon

If no tracefile is generated, check into core_dump_dest and run pstack on the corefile if one is generated and send it to me.

Have been using your tools alot recently… For some reason when I use usql I get a disconnect:

SQL*Plus: Release 10.2.0.3.0 – Production on Wed Dec 9 12:20:35 2009

Copyright (c) 1982, 2006, Oracle. All Rights Reserved.

Enter user-name: / as sysdba

Connected to:
Oracle Database 10g Enterprise Edition Release 10.2.0.3.0 – 64bit Production
With the Partitioning, Real Application Clusters, OLAP and Data Mining options

USERNAME INST_NAME HOST_NAME SID SERIAL# VERSION STARTED SPID CPID SADDR PADDR
——————– ———— ————————- —– ——– ———- ——– ————— ————— —————- —————-
SYS UK1 lscxw01p1.ukdatapoint.com 198 4875 10.2.0.3.0 20091208 6222 6154 000000021F578F08 000000021F40E918

SQL> select * from v$version;

BANNER
—————————————————————-
Oracle Database 10g Enterprise Edition Release 10.2.0.3.0 – 64bi
PL/SQL Release 10.2.0.3.0 – Production
CORE 10.2.0.3.0 Production
TNS for Linux: Version 10.2.0.3.0 – Production
NLSRTL Version 10.2.0.3.0 – Production

SQL> !uname -a
Linux lscxw01p1.ukdatapoint.com 2.6.9-67.0.4.ELsmp #1 SMP Fri Jan 18 05:00:00 EST 2008 x86_64 x86_64 x86_64 GNU/Linux

SQL> @usql 243

HASH_VALUE CH# PLAN_HASH SQL_TEXT
———- —– ———- ————————————————————————————————————–
(XMLAGG889115592 /* Formatted on 2009/09/04 12:11 (Formatter Plus v4.8.7) */
(“venue”,ELEMENT
(ea.venue_id AS “venue_id”,
FROM mv_attribute_value
AND activity_id = 301
),te_name = ‘venue_name’) AS “venue_name”
(SELECT XMLAGG
(“title”,
xmlattributes

CH# PARENT_HANDLE OBJECT_HANDLE PARSES H_PARSES EXECUTIONS FETCHES ROWS_PROCESSED LIOS PIOS SORTS CPU_MS ELA_MS USERS_EXECUTING
—– —————- —————- ———- ———- ———- ———- ————– ———- ———- ———- ———- ———- —————
0 0000000206CEF2A0 0000000206AD0200 1 1 5 4 4 32488739 16856 68775 867697.956 1240898.07 1

SQL hash value: 586338503 Cursor address: 0000000206CEF2A0
| Statement first parsed at: 2009-12-09/11:19:26 | 3690 seconds ago

select –+ ordered use_nl(p ps)
*
ERROR at line 1:
ORA-03113: end-of-file on communication channel

ERROR:
ORA-03114: not connected to ORACLE

Have you seen this before?

All other sqls; snapper, sw etc work fine.

Any thoughts,

Regards,

Simon