Re: gc buffer busy waits

From: Riyaj Shamsudeen <riyaj.shamsudeen_at_gmail.com>
Date: Thu, 10 Mar 2011 10:10:40 -0600
Message-ID: <AANLkTi=HY9Ui6FwmiSARtnvAJNpx=DzKywSstWajbFJf_at_mail.gmail.com>



Hello

   gc buffer busy waits simply means that there is a pending request for BL lock on that block and the current session must wait for the request to be complete. It is usually a symptom of concurrency.   You need to review further to understand the concurrency. First thing to do is to find which block in that object is suffering from gc buffer busy wait. If it is a segment header block, then it is probably an issue with freelists setup or ASSM setup. I would think, may be that ( you are not using ASSM tablespace) or (that you are using MSSM tablespace and have not setup freelists properly etc). Again, that would be my guess, and need further review to help you properly.
 If you have access to ASH, you could group by file_id and block_id for the sessions waiting for gc buffer busy waits to see the pattern.

 I blogged about this topic here:
   http://orainternals.wordpress.com/2010/09/27/gc-buffer-busy-waits/ Cheers

Riyaj Shamsudeen
Principal DBA,
Ora!nternals - http://www.orainternals.com - Specialists in Performance, Recovery and EBS11i
Blog: http://orainternals.wordpress.com
Upcoming class: Intensive RAC Webinar in June 2011! OakTable member http://www.oaktable.com
Co-author: "Expert Oracle practices: Oracle Database Administration from the Oak Table" http://www.apress.com/book/view/9781430226680

On Wed, Mar 9, 2011 at 12:47 PM, <lyallbarbour_at_sanfranmail.com> wrote:

> Not sure if this has been asked, but i just don't know what to do with the
> information that ADDM and ASH reports give me.
>
> We have a large table DA.ST_INV_ITEMS_HS, almost 690,000 rows. Big table
> with lots of columns. The table takes up 632Meg. It has only 173 extents.
> 8k block database 10gR2 RAC 2 node cluster. Tons of cluster waits on that
> table. I always thought that that means that one session has queried off
> this table, locked those blocks or extents in memory on one node and a
> session on the other nodes is waiting for those blocks or extents to be
> unlocked.
>
> can someone explain what the ASH report means about Top DB Objects?
>
> Top DB Objects
>
> - With respect to Application, Cluster, User I/O and buffer busy waits
> only.
>
> Object ID% ActivityEvent% EventObject Name (Type)Tablespace 54768 17.04 gc
> buffer busy 8.80 DA.ST_INV_ITEMS_HS (TABLE) USERS gc cr multi block
> request 7.70
>
>
>

--
http://www.freelists.org/webpage/oracle-l
Received on Thu Mar 10 2011 - 10:10:40 CST

Original text of this message