Comments on: latch: cache buffers chains latch contention – a better way for finding the hot block http://blog.tanelpoder.com/2009/08/27/latch-cache-buffers-chains-latch-contention-a-better-way-for-finding-the-hot-block/?utm_source=rss&utm_medium=rss&utm_campaign=latch-cache-buffers-chains-latch-contention-a-better-way-for-finding-the-hot-block Oracle, Exadata, Performance, Troubleshooting - Mobile Life and Productivity. Wed, 08 Feb 2012 08:03:50 +0000 hourly 1 http://wordpress.org/?v=3.3.1 By: Gabriel Fernando Castillo Contreras http://blog.tanelpoder.com/2009/08/27/latch-cache-buffers-chains-latch-contention-a-better-way-for-finding-the-hot-block/comment-page-1/#comment-11370 Gabriel Fernando Castillo Contreras Fri, 27 Jan 2012 23:31:04 +0000 http://blog.tanelpoder.com/?p=417#comment-11370 Hi, I'm trying to download the script bhla , but the link is broken. http://www.tanelpoder.com/files/scripts/bhla.sql can you say me were can i find it. tnxs Hi, I’m trying to download the script bhla , but the link is broken.
http://www.tanelpoder.com/files/scripts/bhla.sql
can you say me were can i find it.
tnxs

]]> By: charles http://blog.tanelpoder.com/2009/08/27/latch-cache-buffers-chains-latch-contention-a-better-way-for-finding-the-hot-block/comment-page-1/#comment-11086 charles Wed, 04 Jan 2012 04:40:28 +0000 http://blog.tanelpoder.com/?p=417#comment-11086 Hi, sorry to bug you, why couldn't I download your scripts? each time, when I clicked on your links, I got not found. Since your scripts are fantastic, I really wish I could take advantage of them. Thanks and wish you have a great day! Hi,
sorry to bug you, why couldn’t I download your scripts? each time, when I clicked on your links, I got not found.
Since your scripts are fantastic, I really wish I could take advantage of them. Thanks and wish you have a great day!

]]> By: Tanel Poder http://blog.tanelpoder.com/2009/08/27/latch-cache-buffers-chains-latch-contention-a-better-way-for-finding-the-hot-block/comment-page-1/#comment-7030 Tanel Poder Thu, 21 Apr 2011 18:50:12 +0000 http://blog.tanelpoder.com/?p=417#comment-7030 <a href="#comment-7029" rel="nofollow">@David Fitzjarrell</a> Check my blog front page (or the link above) you can download the scripts from there @David Fitzjarrell

Check my blog front page (or the link above) you can download the scripts from there

]]> By: David Fitzjarrell http://blog.tanelpoder.com/2009/08/27/latch-cache-buffers-chains-latch-contention-a-better-way-for-finding-the-hot-block/comment-page-1/#comment-7029 David Fitzjarrell Thu, 21 Apr 2011 18:45:05 +0000 http://blog.tanelpoder.com/?p=417#comment-7029 The links to your scripts are broken - I see only a 404 error page. :( The links to your scripts are broken – I see only a 404 error page. :(

]]> By: Tanel Poder http://blog.tanelpoder.com/2009/08/27/latch-cache-buffers-chains-latch-contention-a-better-way-for-finding-the-hot-block/comment-page-1/#comment-6989 Tanel Poder Fri, 15 Apr 2011 20:43:40 +0000 http://blog.tanelpoder.com/?p=417#comment-6989 Please download the scripts from the .zip here for now: http://blog.tanelpoder.com/2011/04/15/new-scripts-tools-and-broken-links/ Please download the scripts from the .zip here for now:

http://blog.tanelpoder.com/2011/04/15/new-scripts-tools-and-broken-links/

]]> By: aliyar http://blog.tanelpoder.com/2009/08/27/latch-cache-buffers-chains-latch-contention-a-better-way-for-finding-the-hot-block/comment-page-1/#comment-6987 aliyar Fri, 15 Apr 2011 17:49:20 +0000 http://blog.tanelpoder.com/?p=417#comment-6987 very useful informations tanel .. thanks a lot .. but getting " page could not be opened" for all the script links .. very useful informations tanel .. thanks a lot ..

but getting ” page could not be opened” for all the script links ..

]]> By: joel garry http://blog.tanelpoder.com/2009/08/27/latch-cache-buffers-chains-latch-contention-a-better-way-for-finding-the-hot-block/comment-page-1/#comment-6552 joel garry Mon, 14 Mar 2011 22:36:11 +0000 http://blog.tanelpoder.com/?p=417#comment-6552 Getting a 404 for bhla.sql link. Getting a 404 for bhla.sql link.

]]> By: fabio http://blog.tanelpoder.com/2009/08/27/latch-cache-buffers-chains-latch-contention-a-better-way-for-finding-the-hot-block/comment-page-1/#comment-5277 fabio Thu, 07 Oct 2010 18:40:19 +0000 http://blog.tanelpoder.com/?p=417#comment-5277 <a href="#comment-5264" rel="nofollow">@Tanel Poder</a> I'll get the info and will post the details. 1) 10.2.0.4 2) OS is solaris 10, the server model is sun fire e25k, wc -l shows 56 cpus 3) There are between 100 and 300 sessions trying to do the update, depending on the load we try to achieve. This question you asked is what I am trying to understand. Neither the vendor or the application folks were able to tell how the column grows. I will do some research using logminer. They should have more tests scheduled soon, will send the snapper updates. Thanks a lot for getting back fabio @Tanel Poder
I’ll get the info and will post the details.
1) 10.2.0.4
2) OS is solaris 10, the server model is sun fire e25k, wc -l shows 56 cpus
3) There are between 100 and 300 sessions trying to do the update, depending on the load we try to achieve.

This question you asked is what I am trying to understand. Neither the vendor or the application folks were able to tell how the column grows. I will do some research using logminer.

They should have more tests scheduled soon, will send the snapper updates.

Thanks a lot for getting back

fabio

]]> By: Tanel Poder http://blog.tanelpoder.com/2009/08/27/latch-cache-buffers-chains-latch-contention-a-better-way-for-finding-the-hot-block/comment-page-1/#comment-5264 Tanel Poder Wed, 06 Oct 2010 18:30:35 +0000 http://blog.tanelpoder.com/?p=417#comment-5264 <a href="#comment-5263" rel="nofollow">@fabio</a> Thanks, few questions: 1) Which exact db version? 2) Which hardware platform? (is it a Sun T5xxx coolthreads server perhaps?) 3) How many concurrent updates/sessions doing these updates? Are you updating column values larger, so that the row grows, doesn't fit into the block and has to be inserted into a new block? Please run snapper on one of the sessions and post the output here (or send by email so I can blog about it later :) to see wee what else is happening during these updates... @fabio

Thanks, few questions:

1) Which exact db version?
2) Which hardware platform? (is it a Sun T5xxx coolthreads server perhaps?)
3) How many concurrent updates/sessions doing these updates?

Are you updating column values larger, so that the row grows, doesn’t fit into the block and has to be inserted into a new block?

Please run snapper on one of the sessions and post the output here (or send by email so I can blog about it later :) to see wee what else is happening during these updates…

]]> By: fabio http://blog.tanelpoder.com/2009/08/27/latch-cache-buffers-chains-latch-contention-a-better-way-for-finding-the-hot-block/comment-page-1/#comment-5263 fabio Wed, 06 Oct 2010 18:25:00 +0000 http://blog.tanelpoder.com/?p=417#comment-5263 tanel, first of all, thanks for all that info you disclosed to us :) i have an situation hitting cache buffer chains for an "update" - which happens quite frequently. the access is already done by rowid. we are able to keep up with around 2 million updates in 15 minutes. a long raw column is updated as well, no indexed columns are updated, typical size of the row is 2.5 kbytes. the tablespace for this table is ASSM. i have tried your technique and the hot block i get is the segment header for this object. usually we see this segment header thing with inserts only. there is no cpu starvation. would you point me to some other doc or post for further investigation? my ideas so far are to increase the pctfree to sparse the blocks and increase further the sga to obtain more chains. thanks, fabio tanel,
first of all, thanks for all that info you disclosed to us :)

i have an situation hitting cache buffer chains for an “update” – which happens quite frequently. the access is already done by rowid. we are able to keep up with around 2 million updates in 15 minutes. a long raw column is updated as well, no indexed columns are updated, typical size of the row is 2.5 kbytes.

the tablespace for this table is ASSM. i have tried your technique and the hot block i get is the segment header for this object. usually we see this segment header thing with inserts only.

there is no cpu starvation.

would you point me to some other doc or post for further investigation? my ideas so far are to increase the pctfree to sparse the blocks and increase further the sga to obtain more chains.

thanks,
fabio

]]>