Search

Top 60 Oracle Blogs

Recent comments

March 2011

Oracle by Example portal now shows 12g

A quick post… which is a bit interesting…

As I was searching for Exadata viewlets in Oracle By Example portal (http://goo.gl/HjshR) I clicked on the “Advanced Search”

and then filtered by Product Family (Database) and then by Product

to my surprise there was an option for “Database 12g”….

Battle: Los Angeles…

For some masochistic reason you decide to watch Independence Day, realize it’s a steaming pile of crap (of almost Twilight proportions) and decide the only way to make the world a better place is to redo the film properly. What you end up with is Battle: Los Angeles. It’s pretty much the same story.

I’m not saying it’s a great sci-fi film, but it’s pretty darn good. I wasn’t holding out much hope during the first few scenes because the person with the camera seemed to be having a fit. On stationary shots where people were just chatting it looked like the camera was sitting on a washing machine during the spin cycle. Once the action kicked in the camera work seemed much more appropriate, even steady at times.

The feel was very much like that of District 9 and Skyline, which is a good thing in my opinion. Dirty and gritty wins out over Hollywood polish for me. The budget was quite a bit higher than those films, but it had some recognizable actors and everything was so much bigger. It still came in at $5,000,000 under the price of Independence Day, and that was made 14 years ago.

There were a few inevitable bits of cheese and gaps in story development, probably from cuts to bring it just under 2 hours, but it still worked pretty well.

I was surprised to see Ne-Yo in the film, but he did an OK job. If Michelle Rodriguez is one tenth as cool in real life as she is in films, she will be mine. Oh yes! She will be mine! It would be great to have a girlfriend who could beat the crap out of guys if they don’t respect my authoritah:)

Cheers

Tim…




Battle: Los Angeles…

For some masochistic reason you decide to watch Independence Day, realize it’s a steaming pile of crap (of almost Twilight proportions) and decide the only way to make the world a better place is to redo the film properly. What you end up with is Battle: Los Angeles. It’s pretty much the same story.

I’m not saying it’s a great sci-fi film, but it’s pretty darn good. I wasn’t holding out much hope during the first few scenes because the person with the camera seemed to be having a fit. On stationary shots where people were just chatting it looked like the camera was sitting on a washing machine during the spin cycle. Once the action kicked in the camera work seemed much more appropriate, even steady at times.

The feel was very much like that of District 9 and Skyline, which is a good thing in my opinion. Dirty and gritty wins out over Hollywood polish for me. The budget was quite a bit higher than those films, but it had some recognizable actors and everything was so much bigger. It still came in at $5,000,000 under the price of Independence Day, and that was made 14 years ago.

There were a few inevitable bits of cheese and gaps in story development, probably from cuts to bring it just under 2 hours, but it still worked pretty well.

I was surprised to see Ne-Yo in the film, but he did an OK job. If Michelle Rodriguez is one tenth as cool in real life as she is in films, she will be mine. Oh yes! She will be mine! It would be great to have a girlfriend who could beat the crap out of guys if they don’t respect my authoritah:)

Cheers

Tim…




Visitors etc.

Well, I’ve finally reached another “landmark” number in the page views – 2,000,000 views (after four and a half years) – so it’s time for a little statistical summary for comparison with the last landmark.

Looking at just the results for the last 12 months, here are some numbers in different categories.

Page Views:

AWR / Statspack 11,871 Updated from time to time
NOT IN 7,981 Published February 2007
dbms_xplan in 10g 6,672 Published November 2006
Simple Scripts 5,899 Updated from time to time
Cartesian Merge Join 5,335 Published December 2006

Referrers (incoming links)

Google Reader 13,296
forums.oracle.com 10,234 The OTN database forum. A large percentage for reading AWR, index rebuilds, or hints.
Google 6,079 80/20 split between the news groups and searches
oaktable.net 3,691 The Oak Table site. Most from the aggregator
jlcomp.demon.co.uk 3,235 My old web page. Most are from the front page
orana.info 2,330 Eddie Awad’s feed aggregator
dbaspot.com 1,849 A newsgroup history site. Most are for “How to read AWR”

Clicks (places my readers go to)

Oak Table home page 1,926
My old web page 1,377
Nigel Noble’s blog 1,961
Oracle’s Statspack paper(pdf) 1,238
Wolfgang Breitling’s home page 1,057
The 10053 treeviewer download 981
Graham Woods ASH presentation 936
Julian Dyke’s presentation page 929

Incoming Searches

There are harder to count, so I haven’t, but to demonstrate this – I’ve listed some of the most popular searches by the number of different variations that got to the blog, but I haven’t summed the related numbers of page views.

“Jonathan Lewis” / Scratchpad 23 variations
Oracle NOT IN 20 variations
SQL Profiles 18 variations
“some topic” + “jonathan lewis” or “scratchpad” 16 variations
Cartesian join 14 variations
free buffer waits 12 variations
dbms_xplan 12 variations
push_subq 11 variations

Real-Time SQL Monitoring… Update…

Real-Time SQL Monitoring got a lot of air-time when it was released in 11g. I remember being wowed in a number of presentations. What I had failed to notice until recently was a quite interesting update in 11gR2. Previously, the reports always looked a lot prettier in Enterprise Manager than they did in the HTML reports produced by the DBMS_SQLTUNE package. In 11gR2, the introduction of an ACTIVE report means you can now get the EM-style report using the DBMS_SQLTUNE package. Of course, if you are always using EM to look at your SQL monitoring reports, this change is of little value, but if you are using DBMS_SQLTUNE to generate the reports you may want to give it a try because the layout is definitely better. I’ve updated my article on the subject to reflect the changes in 11gR2.

Just goes to show, every time you open the manuals something pops out to make you realize how much you don’t know.

Cheers

Tim…




Real-Time SQL Monitoring… Update…

Real-Time SQL Monitoring got a lot of air-time when it was released in 11g. I remember being wowed in a number of presentations. What I had failed to notice until recently was a quite interesting update in 11gR2. Previously, the reports always looked a lot prettier in Enterprise Manager than they did in the HTML reports produced by the DBMS_SQLTUNE package. In 11gR2, the introduction of an ACTIVE report means you can now get the EM-style report using the DBMS_SQLTUNE package. Of course, if you are always using EM to look at your SQL monitoring reports, this change is of little value, but if you are using DBMS_SQLTUNE to generate the reports you may want to give it a try because the layout is definitely better. I’ve updated my article on the subject to reflect the changes in 11gR2.

Just goes to show, every time you open the manuals something pops out to make you realize how much you don’t know.

Cheers

Tim…




Copying the password file for RAC databases

This post is inspired by a recent thread on the oracle-l mailing list. In post “11g RAC orapw file issue- RAC nodes not updated” the fact that the password file is local to the instance has been brought up. In fact, all users with the SYSOPER or SYSDBA role granted are stored in the password file, and changing the account for the SYS user on one instance doesn’t mean the password change is reflected on the other RAC instances. Furthermore, your Data Guard configuration will break as well, since the SYS account is used to log in to the standby database.

On a related note, the change of the sys password for the ASM instance in GRID_HOME will propagate to all cluster nodes automatically, a fact I have first seen mentioned on the Dutch Prutser’s weblog, Harald van Breederode.

Now to get over the annoyance of having to manually copy the new password file to all cluster nodes I have written a small shell script, which I use for all my Linux clusters. It takes the ORACLE_SID of the local instance for input, then works out the corresponding ORACLE_HOME and copies the password file to all instances in the cluster, as listed in the output of olsnodes. The script can deal with separation of duty, i.e. Systems where GRID_HOME is owned by a different owner then the RDBMS ORACLE_HOME. The script is by no means perfect, and could be extended to deal with a more general setup. My assumption is that all cluster nodes have a 1:1 mapping of Oracle instance and ORACLE_SID, for example instance PROD1 will be hosted on the first cluster node, prodnode1.

The script is shown below, it’s been written and tested on Linux:

#!/bin/bash

# A small and simple script to copy a password file
# to all nodes of a cluster
# This works for me, it doesn't necessarily work for you,
# and the script is provided "as is"-I will not take
# responsibility for its operation and it comes with no
# warrenty of any sorts
#
# Martin Bach 2011
#
# You are free to use the script as you feel fit, but please
# retain the reference to the author.
#
# Usage: requires the local ORACLE_SID as a parameter.
# requires the ORACLE_SID or DBNAME to be in oratab

ORACLE_SID=$1
[[ $ORACLE_SID == "" ]] && {
 echo usage `basename $0` ORACLE_SID
 exit 1
}

#### TUNEABLES

# change to /var/opt/oracle/oratab for Solaris
ORATAB=/etc/oratab
GRID_HOME=/u01/crs/11.2.0.2

#### this section doesn't normally have to be changed

DBNAME=${ORACLE_SID%*[0-9]}
ORACLE_HOME=`grep $DBNAME $ORATAB | awk -F":" '{print $2}'`
[[ $ORACLE_HOME == "" ]] && {
 echo cannot find ORACLE_HOME for database $DBNAME in $ORATAB
 exit 2
}

cd $ORACLE_HOME/dbs
cp -v orapw$ORACLE_SID /tmp
INST=1

echo starting copy of passwordfile
for NODE in `$GRID_HOME/bin/olsnodes`; do
 echo copying orapw$ORACLE_SID to $NODE as orapw${DBNAME}${INST}
 scp orapw$ORACLE_SID $NODE:${ORACLE_HOME}/dbs/orapw${DBNAME}${INST}
 INST=$(( $INST + 1))
done

It’s fairly straight forward, we first get the ORACLE_SID and use this to get the ORACLE_HOME for the database.  The GRID_HOME has to be hard coded to keep it compatible with < 11.2 database where you could have a CRS_HOME different from the ASM_HOME. For Oracle < 11.2, you need to set the GRID_HOME variable to your Clusterware home.

The DBNAME is the $ORACLE_SID without trailing number, which I need to work out the SIDs of the other cluster nodes. Before copying the password file from the local node to all cluster nodes a copy is taken to /tmp, just in case.

The main logic is in the loop provided by the output of olsnodes, and the local password file is copied across all cluster nodes.

Feel free to use at your own risk, and modify/distribute as needed. This works well for me, especially across the 8 node cluster.

Moats

The (M)other (O)f (A)ll (T)uning (S)cripts. A tuning and diagnostics utility for SQL*Plus co-developed with Tanel Poder. Includes active session sampling and a TOP-style utility for displaying instance activity at regular refresh intervals. Supports versions from 10g Release 2 onwards. September 2010 (updated March 2011)

I can relate...

It is not often a simple picture and caption will make me laugh out loud - but this did. I feel like that some days... Most days now I guess :)

Update, things of interest and a couple of blogs to check out

tech.E2SN secret hacking session on Tuesday 22nd March:

Just in case you missed it – there’s still chance to sign up to my tomorrow’s ORA-4031 and shared pool hacking session. I initially planned to limit the attendees to 100 per event (as the limited GotoWebinar package is cheaper that way) but over 100 people had signed up for the US event on the day of announcement, even before it was 8am in California, so I figured I should invest a bit more and allow more people attend. So far over 500 people have signed up (total for both events). If you haven’t done so, you can sign up here:

Advanced Oracle Troubleshooting online seminar Deep Dives 1-5  on 11-15 April:

The next AOT deep dives (1-5) will start in 3 weeks, on 11-15 April. (and 6-10 will be on 9-13 May).

Check the details here:

Blogs to check out:

Andrey Nikolaev has done some serious low-level research on Oracle latches and KGX mutexes and he also presented his work this year at Hotsos Symposium (I missed his session as I was stuck in JFK instead of attending the conference on that day):

Porus Havewala is quite a Grid Control and OEM enthusiast. If you are into OEM & GC, check out his blog:

Future events:

  • I will be speaking at the UKOUG Exadata Special Event on 18th April
  • I will announce some more Virtual Conferences pretty soon!!! Very interesting topics and good speakers – including (but not limited to) some serious Exadata technical contents!

Share