Tunning IO - AIX - RAC
Anybody with implementations experience on RAC 10g in AIX 5.2 with ASM, could spend some tips for tuning of IO for AIX?
I look at several papers mentioning setting Asynchronous I / O, but not change slow IO performance.
Thanks.
Hi Donald, Thanks for your reply!
I am using Raw Device with ASM --> mknod /dev/ASM_Disk1 c 20 11
Create a Disk Group DATA with external redundancy on ASM_DISK1.
CREATE TABLESPACE TBSDAT_PART_LANCAMENT_2007_N DATAFILE
'+DATA' SIZE 650M AUTOEXTEND ON NEXT 100M MAXSIZE 2500M
LOGGING
ONLINE
PERMANENT
EXTENT MANAGEMENT LOCAL UNIFORM SIZE 100M
BLOCKSIZE 8K
SEGMENT SPACE MANAGEMENT AUTO
FLASHBACK ON;
CREATE TABLE TB_TESTE
CD_ANO NUMBER(4) NOT NULL,
CD_ATRIBUTO VARCHAR2(30 BYTE),
CD_LANCAMENTO NUMBER NOT NULL,
TABLESPACE TBSDAT_PART_LANCAMENT_2007_N
LOGGING
PARTITION BY RANGE (CD_ANO)
PARTITION LANCAMENTO_P2007 VALUES LESS THAN (2008)
LOGGING
NOCOMPRESS
TABLESPACE TBSDAT_PART_LANCAMENT_2007_N)
NOCOMPRESS
NOCACHE
PARALLEL ( DEGREE 4 INSTANCES 1 )
MONITORING;
Insert 10 records in table.
SELECT * FROM TB_TESTE WHERE CD_ANO = 2007
sar -u 1 100
09:19:34 %usr %sys %wio %idle
09:19:35 4 2 90 10
09:19:36 4 3 92 08
09:19:37 4 3 90 10
Takes 50 seconds to return!!
How can I resolve this??
Thank's
Similar Messages
-
Hi all,
Currently i am working as a core dba . My company want to put me on different project which is running
on rac environment IBM-AIX servers . Now some queries are
taking more than 2 hours and reports are taking more than 8 hours(cognos). My work is to tune the database and sql statements and finding areas which needs to tune.
In rac environment how to find log running sql or packages/procedures. can i use V$sql here ..?
Like what first steps i need to take ..where i can find bottleneck ..like for wait events i will go for
V$ wait views..etc.
I need suggestions.
ThanksI find EM and AWR reports very useful in tuning both single instances and RAC databases. But you can also look into the various v$-views. In a RAC environment, you can also query most of the v$-views with gv$. Those are global views spanning all instances. So gv$session will show all sessions on all nodes while v$session only reports on the node your session is connected to. Same thing for gv$sql, gv$session_longops and so on
-
Hi Gurus,
I have been asked to take an interview for the post of ORACLE DBA.
This job profile is to support RAC databases.
I will be questioning DBA's on the following.
INSTALL AND CONFIGURE RAC
TAF
BACKUP AND RECOVERY OF RAC database.
But my main question is how I can judge the DBA about his RAC experience.
Regards,
RiyadhI am taking a shot at the questions:-
Question:Tune the following RAC DATABASE (DBNAME=PROD) which is 3 >node RAC.
PROD1 PROD2 PROD3
CPU 8 CPU 15 CPU 8
32 GB RAM 12 GB RAM 16 GB RAMWhat are you looking for here? What tuning information do you expect?
It is a 3 node cluster with different hardware configuration running RAC.
I would put 20% of the memory for Oracle in each node. So that would mean that the SGA is different in each of the nodes.
Also since the CPU's are different PROD2 can have more number of max number of processes as compared to the rest of them.
But as I said this is just configuration, this is not tuning. Question is not clear.
Question: Write a RMAN script for taking backup of the database including arch > log files.
ARCH LOG LOCATION = /ORACLE/ARCH on each machine.run
allocate channel sbt_table .... ;
backup database ;
backup archive log all delete all input;
Question:Write a sample script for RMAN for the recovery if all the instance are > down.(First explain the procedure how you will restore)Bring all nodes down.
Start one Node
Restore all datafiles and archive logs.
Recover 1 Node.
Open the database.
bring other nodes up.
Confirm that all nodes are operational.
Some questions from my side:-
Q. Clien'ts are performing some operation and suddenly one of the datafile is experiencing problem what do you do?
The cluster is a two node one.
A. Bring the datafile offline recover the datafile.
Q. How can you connect to a specific node in a RAC environment?
A. tnsnames.ora ensure that you have INSTANCE_NAME specified in it.
Q.What is the difference between a OS cluster and a rac cluster?
Look for my post above in this forum.
Q.What is gsd?
Q. What is cachefusion?
Q. What happens when a DML is issued in a RAC environment, how are requests for common buffers handled in a RAC environment. -
Can we implement 10g RAC without ASM
Hi,
OS - AIX 5.3
DB - 10g R2
Clusterware - 10g R2
Can we implement 10g RAC using RAW files without ASM ?
If yes then, do we have to install HACMP if we use RAW files only ?
Regards,
Amityou really do NOT need hacmp
check the AIX RAC Installation Guide (Chapter 3 explains the different storage solutions for AIX)
Here is the overview
Storage Option File Types Supported
OCR and Voting Disk Oracle Software Database Recovery
Automatic Storage Management No No Yes Yes
General Parallel File System (GPFS) Yes Yes Yes Yes
Local storage No Yes No No
Raw Logical Volumes Managed by HACMP Yes No Yes Yes -
Good day!
I have 27 schemas (which contains 1000+ tables each) and I needed to replicate all of it in our DR machine, my question, is it possible to create multiple initial load extract for me to be able to finish initial loading in a small amount of time?
Thanks!!
Regards,
MelaI'm having this kind of problem.
I need to migrate database from Solaris with raw devices to AIX RAC with ASM.
As it is a production server and it cannot spend to much time down to transfer huge data using expdp from one server to another, the only way I found to do it is using GoldenGate.
Can somebody help me?
Regards, -
Disable automatic Agent restart
I need to temporarily disable the agent restart when a RAC node is bounced.
Can someone tell me how the agent is configured for automatic restart on an AIX RAC node?Yeah, I knew about the watchdog.
I'm talking about server restarts though.
I have not specifically configured the em agent (10.2.0.2) to restart when the cluster node restarts; however, it does restart on it's own. This is also the case with the non-RAC em agent installs. I could not locate anything in /etc/init.d for this purpose. I seem to recall reading something about gcstart or something like that, but I could not locate the document.
Thanks. -
hello all,
this is 10g on AIX(RAC). i have been trying to use a index on a query forcefully by using a HINT ..here is the query.
SELECT m.* FROM law.messages m inner join law.log_events l on l.bod_id = m.bod_id and l.type = 'LogEventException' WHERE ((m.source_date_time between
to_date('2008-10-01 14:58:00', 'yyyy/mm/dd hh24:mi:ss') and
to_date('2008-10-02 14:58:00', 'yyyy/mm/dd hh24:mi:ss')) and m.bod_id!='unknown') ORDER BY m.source_date_time desc;
and my question it still not using an index i want it to use. the index name is IDX_MSG_SOURCE_DATE_TIME .....its on message table...wat am i doing wrong here?? this is wat i using the hint as...
SELECT /*+ index (message idx_msg_source_date_time)*/ m.* FROM law.messages m inner join law.log_events l on l.bod_id = m.bod_id and l.type = 'LogEventException' WHERE ((m.source_date_time between
to_date('2008-10-01 14:58:00', 'yyyy/mm/dd hh24:mi:ss') and
to_date('2008-10-02 14:58:00', 'yyyy/mm/dd hh24:mi:ss')) and m.bod_id!='unknown') ORDER BY m.source_date_time desc;
i look at the explain plan and everything...it dosent use the index i want it to use..so wat am i doing wrong?? i did drop and re-created the index...gatherd stats for the index...tables and whole schema as well...still no results...
the query dose pick up the index or shall i say use a index on log_Event table ....but dont use index on message table...which is making this query run forever...
Edited by: user630084 on Oct 14, 2008 10:33 AM
Edited by: user630084 on Oct 14, 2008 10:36 AMHi,
check the EP of below sql. it should use the index.
SELECT /*+ index (m idx_msg_source_date_time)*/ m.* FROM law.messages m inner join law.log_events l on l.bod_id = m.bod_id and l.type = 'LogEventException' WHERE ((m.source_date_time between
to_date('2008-10-01 14:58:00', 'yyyy/mm/dd hh24:mi:ss') and
to_date('2008-10-02 14:58:00', 'yyyy/mm/dd hh24:mi:ss')) and m.bod_id!='unknown') ORDER BY m.source_date_time desc;
Thanks,
krishna B -
10G RAC on AIX without ASM & HACMP.
Hi,
I want to install ORACLE 10G RAC on AIX but I dont want ot use ASM or HACMP/GPFC.
Just want to install with CRS and RAW devices.Is it possible to install 10g RAC with CRS and RAW devices.
What is the best approach for this.
Regards
MMUWe're running CRS on AIX. No ASM or HACMP.
We do have the db datafiles on GPFS. We've also got one voting disk on GPFS and two on raw devices. -
11gR2 RAC installation in AIX fails while running root.sh on node2
Hi
We are in the process on installing 11gR2 RAC on AIX 6.1
But the installation is failing on node 2while running root.sh with the following error.
DiskGroup DATA creation failed with the following message:
ORA-15018: diskgroup cannot be created
ORA-15017: diskgroup "DATA" cannot be mounted
ORA-15003: diskgroup "DATA" already mounted in another lock name space
Configuration of ASM failed, see logs for details
Did not succssfully configure and start ASM
CRS-2500: Cannot stop resource 'ora.crsd' as it is not running
CRS-4000: Command Stop failed, or completed with errors.
Command return code of 1 (256) from command: XXXXX/grid/bin/crsctl stop resource ora.crsd -init
Stop of resource "ora.crsd -init" failed
Failed to stop CRSD
Please help
RegardsHI,
the second node (second root.sh) should not want to "create" the diskgroup data, since it is already there.
This pretty much sounds like you tried to start both root.sh (on the first node and on the second node) at the same time, not waiting for it to finish on the first node.
It is important, that before you start the root.sh on the second node, it has to be finished successfully on the first node for the cluster to update the information.
Sebastian -
How to tune buffer busy waits in RAC
Hi,
In our environment we se lot of buffer busy waits.
How can I tune that.
we are using 9i RAC on HACMP.
Regards
MMUThere could be several reasons for buffer busy waits, however the common reason is high logical reads..
Normally they relate to bad SQL, you can try to identify those bad queries using a 10046 trace at level 8 or 12 and then tkrprof the trace file and look at the output.
In a RAC environment you will see global buffer busy waits which again relates to buffer busy waits at the instance level. -
rac waits after restarting database.
buffer busy global CR 6,083,927 1,386,593 48.20
global cache cr request 40,081,544 960,533 33.39
buffer busy waits 5,081,718 428,610 14.90
PX Deq: Signal ACK 17,248 42,099 1.46
db file sequential read 3,176,639 20,033 .70
please state where is the problemHi,
GPFS is not free, and needs to be purchased.
ONLY GPFS 2.3 is certified with RAC as for today.
Certification for 3.1 ion under way.
About 9iRAC and HACMP on AIX 5.3, ONLY HACMP 5.1,5.2 and 5.3 are supported, not yet HACMP 5.4. For info, 10gRAC is only supported with HACMP 5.1 and 5.2.
To find out how to setup HACMP for a 9iRAC installation, you can use the following document, available on http://www.oracleracsig.org :
CookBook V3.2 - Oracle 9iRAC on IBM@server pSeries running AIX5L
(Step-by-step installation guide for 9iRAC on AIX5L. Whith HACMP for database on concurrent raw devices And IBM GPFS for database on shared cluster filesystem. Including HACMP and GPFS setup.
The document covers HACMP 5.1 or 5.2, not 5.3 release, but it should be nearly the same with some screen changes.
Point which is different with the cookbook for now, it's the way to create the lv (logical volume), you should use the "-T O" options for it.
Hope it will help.
Regards.
Fred. -
Hi,
I have a RACsystem with 2 nodes (oracle 10.2.0.1.0, Linux RedHat 4)
I have a simple query that take a long time but in the same server with a single database it's fast!!??
here is the query:
select count(*) as y0_, this_.NOTIFICATION_TYPE as y1_
from NOTIFICATION this_
where this_.OPCO_ID=1
and this_.SENTDATE is null
and this_.UPDATED_TIME is null
and this_.CREATED_TIME>=sysdate
and this_.CREATED_TIME<=sysdate
group by this_.NOTIFICATION_TYPE
order by this_.NOTIFICATION_TYPE asc
With AWR I can see that I lost a long time in gc_buffer_busy_wait!!
NB:I have a blob in my table
How I can solve my problem?
Message was edited by:
vittelYou have hit one of the many features of RAC. On a single instance database a piece of data is either in its SGA in local memory, or out on disk. On a multi-instance RAC database a piece of data can now either be in the local SGA memory, or the remote SGA memory of the other node, or out on disk.
Given that Oracle assumes that it is quicker to go across the network interconnect between the 2 RAC nodes to retrieve the copy of the data from the other SGA than it is to go all the way out to disk to read that data page, Oracle RAC is biased to using copies of data blocks in the SGAs of the other RAC nodes.
You cannot stop this. Oracle RAC is doing as it is intended to. Either tune and improve the performance of the interconnect between your 2 nodes. Are you using 100 baseT or 1000 baseT or something even faster? It is the latency that is the issue, not the bandwidth.
Or stop using RAC, and go back to just one database instance.
John -
The CSSD does not start automatically on Non-RAC - AIX 5L
Hi all,
Database: Oracle 10.2.0.4
O.S AIX 5.3 TL 10
I am facing a problem with the CSSD Non-RAC/Clusterware.
When we restarted the server the service CSSD does not start automatically and there is no relevant messages in the log files of CSSD.
Running the "localconfig reset $ORACLE_HOME", execute successfully but does not start the CSSD.
Running "/etc/init.cssd start" - does not work, nothing happens when we run it
I manually ran "/etc/ init.cssd run" and the service CSSD (ossd.bin) was started without errors.
No errors on errpt from AIX.
I've opened a SR but I am waiting for response from Oracle.Hi,
I believe this problem is not with Oracle, but in AIX.It may be that one process that is starting on boot is holding the other process. You need to identify what is process/application.
Check the process on inittab that are already active and which are inactive they should be active.
One clue: Is very common process that is initiated in wait mode hold the boot server.
wait
When the init command enters the run level that matches the entry's run level,
start the process and wait for its termination.
All subsequent reads of the /etc/inittab file while the init
command is in the same run level will cause the init command to ignore this entry.http://publib.boulder.ibm.com/infocenter/aix/v6r1/index.jsp?topic=/com.ibm.aix.files/doc/aixfiles/inittab.htm
Regards,
Levi Pereira -
Question about RAC 10g on server running AIX and domain name changes
We have several 10g clusters running on IBM p-Series AIX servers, and our company is changing the domain name for our network. I am concerned that if we do not change the domain names in the TNSNAMES.ora file and also the /etc/hosts file on each server, the cluster will no longer function. We are using CRS (version 10.2.0.3) and ASM within each cluster.
Does anyone know where I can find information about changing the domain name that was set up as part of the CRS and database installs for RAC? Does anyone have experience doing this, and would be willing to share a few insights? Thanks in advance for any advice you can give!You may be able to use the VIPCA utility for RAC to change or modify the public and private settings for your domain names on AIX. Also, the Oracle 10g DBCA "Database Configuration Assistant" should be of use to change your settings for your RAC environment.
Also, see here for a good IBM Redbook that covers Oracle 10g RAC on AIX platforms:
http://www.redbooks.ibm.com/redbooks/pdfs/sg247541.pdf
Regards,
Ben Prusinski
http://oracle-magician.blogspot.com/
Regards,
Ben Prusinski
http://oracle-magician.blogspot.com/ -
RAC Installation on AIX 5.3 oracle(10.2.0.4.0)
Hi,
I am planning to setup RAC on AIX 5.3 in oracle 10.2.0.4.0
Please provide me any document for step by step installation of RAC on AIX 5.3 in oracle 10.2.0.4.0.
If possible provide me metalink document ID.
ThanksHi user;
Please check below notes and links:
IBM General Parallel File System (GPFS) and Oracle RAC [ID 302806.1]
RAC Assurance Support Team: RAC Starter Kit and Best Practices (AIX) [ID 811293.1]
RAC: Frequently Asked Questions [ID 220970.1]
I also suggest check below googling which has very great notes:
http://www.google.com.tr/search?hl=tr&source=hp&q=RAC+on+AIX&aq=f&aqi=&aql=&oq=&gs_rfai=
Regard
Helios
Maybe you are looking for
-
How to change the semantic color of entire rows in a table.
Hi, in the past I already managed to change the semantic color of a cell by a condition (for example all the cells with negative number -> I set negative semantic color) My question is how can I change the semantic color of the entire row where this
-
How do I add a field in my FormCentral application for applicants to attach files?
How do I add a field in my FormCentral application for applicants to attach files?
-
how can i buy iphone if iam not a canadian but i am in toronto on visit visa and i want to buy an iphone 5 today coz i am going back to Pakistan tomorrow ! please help me fast !!!!
-
Macbook Pro Retina trackpad help.
Hi I just got my mbpr 15'' ( 2013 model) probably less than a month ago and I must say I love it. However lately the track pad has been erractic and jumping here and there, right clicking when you click the left side, sometimes left clicking when you
-
Check the attached image, it says connect a USB cable to upgrade the software, but Nokia 112 does not have the option of USB cable, how do we connect a USB? lol , isn't it funny? if u do not support upgrade of Nokia 112 then why u have allowed the ne