RE: enq: TX - index contention
- From: "Powell, Mark D" <mark.powell@xxxxxxx>
- To: "'Les.Hollis@xxxxxx'" <Les.Hollis@xxxxxx>, oracle-l@xxxxxxxxxxxxx
- Date: Tue, 22 Mar 2005 14:41:12 -0500
I withdraw this post. I think I misunderstood his last point. I will have
to run a few more block dumps and repost with any necessary corrections.
-----Original Message-----
From: oracle-l-bounce@xxxxxxxxxxxxx [mailto:oracle-l-bounce@xxxxxxxxxxxxx]
On Behalf Of Powell, Mark D
Sent: Tuesday, March 22, 2005 2:22 PM
To: 'Les.Hollis@xxxxxx'; oracle-l@xxxxxxxxxxxxx
Subject: RE: enq: TX - index contention
Les, I am not sure how you came to your conclusion that the number of blocks
in the allocation request affects the number of ITL lists allocated, but it
is wrong.
My initial test showed I requested 1 ITL and 4 blocks. I got 4 blocks with
two ITL slots in them. I repeated my test using a tablespace with a uniform
extent size of 256K. I asked for 16k but got 32 8K blocks with two ITL
slots in each.
Errors in the Oracle documentation are not uncommon. I checked the 10g SQL
manual for the physical attributes clause and it still claims 1 is the
default except for two exceptions: clusters and indexes.
But anybody who looks at the block dump knows better. If you still do not
believe the information that I, Gopalakrishnan, or others posted then I do
think there is anything anyone can say that will change your mind until
Oracle updates the documentation.
-- Mark D Powell --
-----Original Message-----
From: oracle-l-bounce@xxxxxxxxxxxxx [mailto:oracle-l-bounce@xxxxxxxxxxxxx]
On Behalf Of Hollis, Les
Sent: Tuesday, March 22, 2005 1:32 PM
To: Powell, Mark D; oracle-l@xxxxxxxxxxxxx
Subject: RE: enq: TX - index contention
I think the confusion here IS that Oracle allocates 1 (ONE) initrans
(ITL) for EACH BLOCK that is allocated to the table. By default, oracle
allocates 2 BLOCKS (Database blocks) for every table (5 for every index)
that gets created. So, I can well see where you would/could get 2 ITL for a
newly created table based on the 2 data blocks allocated.=20
Now, if you were using LMT, I'm sure that number would be greater based on
the larger number of DB blocks for the uniform size of the LMT.
However, it still only allocates ONE ILT for each block unless you specify
OTHERWISE. Hence the documentation WOULD be right.
-----Original Message-----
From: Powell, Mark D [mailto:mark.powell@xxxxxxx]=20
Sent: Tuesday, March 22, 2005 12:11 PM
To: Hollis, Les; oracle-l@xxxxxxxxxxxxx
Subject: RE: enq: TX - index contention
OK, there appears to be some question to how many ITL lists Oracle
allocates. I made the statement that Oracle allocates 2 probably starting
with version 9.
Ran on Oracle version 9.2.0.5 running on AIX 5.3
UT1 >
UT1 > drop table marktest;
Table dropped.
UT1 > create table marktest (
2 fld1 varchar2(10)
3 ,fld2 number
4 ,fld3 date
5 )
6 tablespace usr
7 initrans 1
8 /
Table created.
-- I specified 1 transaction work list and dictionary shows 1
UT1 > select ini_trans from user_tables where table_name =3D 'MARKTEST'
2 /
INI_TRANS
----------
1
UT1 > select file_id, block_id, blocks
2 from dba_extents
3 where segment_name =3D 'MARKTEST'
4 /
FILE_ID BLOCK_ID BLOCKS
---------- ---------- ----------
10 18 4
UT1 > spool off
-- However looking at the block dump you see 2 are allocated.
*** 2005-03-22 13:01:24.699
*** SESSION ID:(22.4090) 2005-03-22 13:01:24.693 Start dump data blocks tsn:
5 file#: 10 minblk 19 maxblk 19 buffer tsn: 5 rdba: 0x02800013 (10/19)
scn: 0x0001.f3774a80 seq: 0x01 flg: 0x06 tail: 0x4a800601
frmt: 0x02 chkval: 0x7458 type: 0x06=3Dtrans data Block header dump:
0x02800013 Object id on Block? Y
seg/obj: 0x7a0d csc: 0x01.ea1081a9 itc: 2 flg: O typ: 1 - DATA
fsl: 0 fnx: 0x0 ver: 0x01
Itl Xid Uba Flag Lck Scn/Fsc
0x01 0x0005.01c.000227ae 0x0140264e.1310.18 --U- 1 fsc
0x0000.f3774a80
0x02 0x0005.01a.00022492 0x01403ca4.1304.01 C--- 0 scn
0x0001.ea108193
data_block_dump,data header at 0x11027205c
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
<snip>
-----Original Message-----
From: Hollis, Les [mailto:Les.Hollis@xxxxxx]=20
Sent: Tuesday, March 22, 2005 12:48 PM
To: Hollis, Les; mark.powell@xxxxxxx; oracle-l@xxxxxxxxxxxxx
Subject: RE: enq: TX - index contention
Addendum.....this is cut from the Database Fundamentals II training course
from Oracle Education....Power Point Slides
INITRANS: Guarantees a minimum level of concurrency. It defaults to 1 for a
data segment and 2 for an index segment, guarantees a minimum level of
concurrency. For example, if set to 3, INITRANS ensures that at least three
transactions can concurrently make changes to the block. If necessary,
additional transaction slots can be allocated from the free space in the
block to permit more concurrent transactions to modify rows in the block.
True in both the 9.0.1 release documentation AND in the 9.2.0.x
Documentation.
-----Original Message-----
From: Hollis, Les
Sent: Tuesday, March 22, 2005 11:32 AM
To: 'mark.powell@xxxxxxx'; oracle-l@xxxxxxxxxxxxx
Subject: RE: enq: TX - index contention
I think not.
SQL> create table junky1 (name varchar2(5));
Table created.
SQL> select ini_trans from user_tables where table_name =3D 'JUNKY1';
INI_TRANS
----------
1
-----Original Message-----
From: oracle-l-bounce@xxxxxxxxxxxxx
[mailto:oracle-l-bounce@xxxxxxxxxxxxx] On Behalf Of Powell, Mark D
Sent: Tuesday, March 22, 2005 10:19 AM
To: oracle-l@xxxxxxxxxxxxx
Subject: RE: enq: TX - index contention
I believe that the default number of initrans for a table has been raised to
two as of version 9.0 =20
-- Mark D Powell --
-----Original Message-----
From: oracle-l-bounce@xxxxxxxxxxxxx
[mailto:oracle-l-bounce@xxxxxxxxxxxxx]
On Behalf Of K Gopalakrishnan
Sent: Tuesday, March 22, 2005 10:10 AM
To: stalinsk@xxxxxxxxx
Cc: oracle-l@xxxxxxxxxxxxx
Subject: Re: enq: TX - index contention
Stalin:
TX enqueue waits with LMODE 4 can happen for the following reasons.
Quoting from 'Oracle Wait Interface'
"A wait for the TX enqueue in mode 4 is normally due to one of the following
reasons:
ITL (interested transaction list) shortage Unique key enforcement Bitmap
index entry
Here, we will talk about the ITL, which is a transaction slot in a data
block. The initial number of ITL slots is defined by the INITRANS clause and
is limited by the MAXTRANS clause. By default, a table has
1 ITL and an index has 2 ITLs. Each ITL takes up 24 bytes and contains the
transaction ID in the format of USN.SLOT#.WRAP#. Every DML transaction needs
to acquire its own ITL space within a block before data can be manipulated.
Contention for ITL occurs when all the available ITLs within a block are
currently in use and there is not enough space in the PCTFREE area for
Oracle to dynamically allocate a new ITL slot. In this case, the session
will wait until one of the transactions is committed or rolled back, and it
will reuse that ITL slot. ITL is like a building parking space. Everyone who
drives to the building needs a parking space. If the parking lot is full,
you have to circle the lot until someone leaves the building. "
So I would expect any of the above conditions in your case? You may
want to query the V$segment_statistics for ITL waits to eliminate the ITL
issue. If you have bitmap indexes, that should be one of the casues for
those excesive waits. The wait time is close to 3 seconds and I suspect that
could be because of the bitmap index issue, --=20
Best Regards,
K Gopalakrishnan
Co-Author: Oracle Wait Interface, Oracle Press 2004
http://www.amazon.com/exec/obidos/tg/detail/-/007222729X/
--
--
//www.freelists.org/webpage/oracle-l
--
//www.freelists.org/webpage/oracle-l
Other related posts:
