Tuesday, June 10, 2014

OUG Finland (Harmony) 2014

I'm back a few days now after an eventful OUG Finland Conference. It was a great 2 days of Oracle techie stuff in one of the best conference locations in the world.

OUGF view

The conference kind of started on the Wednesday. It seemed like most of the speakers from across Europe and USA were getting into Helsinki around lunch time. Heli had arranged to be a tour guide for the afternoon and took us to see some of the sights around Helsinki, ending up at dinner at a Viking restaurant. At this we got to meet up with some of the other speakers.

We got to try the local Tar ice-cream. Let's just say it was an experience :-)

2014 06 04 21 52 23

The conference kicked off on the Thursday morning at a nature reserve (Haltia) about 30 minutes outside of Helsinki. I'm sure you have seem some of the photos that all of us were sharing on twitter. Heli did the opening welcome and got to show off how good her English is and how easily she can switch between Finnish and English (you had to be there). Each day consisted of 6 tracks and we also had some keynotes too. I think the highlight for most people on the first day was the keynote by James Morle. If you were not there, then hopefully he will be invited to give it at other conferences around the world.

Google Made Me Stupid

I did my usual thing of tweeting throughout the conference along with most of the other presenters, ACEs and ACE Directors.

My first presentation of the conference was my Sentiment Analysis using Oracle Data Mining. I had a decent enough attendance at this considering the glorious sunshine we had and also who was also presenting at the same time.

That evening there was the conference BBQ and we were entertained by the a band consisting of Oracle Finland employees. When all of the entertainment was finished it was time for some people to get the bus back to their hotel in Helsinki. But for 15 of us we headed off for an overnight camping in the local forest. Many thanks to Heli and Ann for arranging this and Ollie for setting up the camp and looking after us. The camp was a bit of fun but the mosquitoes were a bit of a issue for me. Lets just say 'They love me'

2014 06 06 06 58 072014 06 05 23 07 062014 06 05 23 06 37

Yes the safety briefing did talk about bears

2014 06 05 22 25 15

The next morning we got back in to the conference centre. After the keynote from Graham Wood (who was one of the campers), I had my second presentation on using ORE and how to get started with it. I was very surprised with the attendance, in that there was perhaps twice the number at it that I was hoping to have. This presentation ran on a little longer than hoped and it is more suited to a 1 hour slot than a 45 minute slot.

Me presenting ougf14

After that my presenting duties were over and I got to enjoy many of the presentations during the rest of the day. One of these was the Standard Edition Round table organised by Philippe Fierens, Jan Karrmans and Ann Sjokvit. There was lots of great discussion at this round table and I really hope they get to host this at other conferences.

Did I mention I got to meet Chris Date. Did I mention I got to meet Chris Date :-)

Me and Chris Date

As the conference came to a close the committee made sure that everyone had a small gift before going home. For all of those that were left waiting for the bus back into Helsinki there was one last (or several) jumps to do. Again the photos are on twitter, etc.

This was a great conference that had attendees (not just speakers) from a number of countries across Europe. I enjoyed the conference and getting to meet some old and new friends. Hopefully next year I will be able to fit OUG Finland into my calendar and so should you.

As Heli says "I'm Finnish and this is the end"

Monday, June 2, 2014


In ORE there are a number ways to get you R scripts to run in parallel in the database. One way is to enable the Parallel option in ORE. This is what will be shown in this post. There are other methods of running various ORE commands/scripts in parallel. With these the scripts are divided out and several parallel R processes are started on the server.

But what if you want to use the database parallel feature on some of your ORE other commands?

Why would you want to do this?

Well the main answer is that you might want to use the parallel option of the database for the creation on objects (tables etc) and for selecting and manipulating the data in the database.

How can you enable your ORE connection to use the in-database parallel feature?

ORE 1.4 has a new option that enables the parallel option for your ORE connection in the database. This option is called ore.parallel.

When you enable or set the ore.parallel option, it seems to be the equivalent of running the following:




The exact details is a little unclear, but it seems to be above commands.

The following commands illustrates some options for using the ore.parallel option.

> #

> # Check to see if the ore.parallel is enabled for your ORE connection

> options("ore.parallel")



The NULL returned value tells us that your ORE connections does not have the Parallel option enabled. If the schema had Parallel enabled by default then we would have have a response of TRUE.

The following command turns on the Parallel option for your ORE connection / schema.

> options("ore.parallel" = TRUE)

> options("ore.parallel")


[1] TRUE

When the Parallel option is enabled (TRUE above) the database will use the degree of parallel that is set as default for the schema or the degree of parallel that is defined for the table when it is being used in your ORE commands.

You can changed the degree of parallelism by passing the required degree as a value to the ore.parallel command. In the following, the degree of parallelism is set to 8. We then as ORE what the degree is set to and it tells us that it is 8. So it was set correctly.

> options("ore.parallel" = 8)

> options("ore.parallel")


[1] 8

Monday, May 26, 2014

Oracle R Enterprise (ORE) Tasks for the Oracle DBA

In previous posts I gave the steps required to install Oracle R Enterprise on your Database server and your client machine.

One of the steps that I gave was the initial set of Database privileges that the DB needed to give to the RQUSER. The RQUSER is a little bit like the SCOTT/TIGER schema in the Oracle Database. Setting up the RQUSER as part of the installation process allows you to test that you can connect to the database using ORE and that you can issue some ORE commands.

After the initial testing of the ORE install you might consider locking this RQUSER schema or dropping it from the Database.

So when a new ORE user wants access to the database what steps does the DBA have to perform.

  1. Create a new schema for the user
  2. Grant the new schema the standard set of privileges to connect to the DB, create objects, etc.
  3. Create any data sets in their schema
  4. Create any views to data that exists in other schemas (and grant the necessary privileges, etc

Now we get onto the ORE specific privileges. The following are the minimum required for your user to be able to connect to their Oracle schema using ORE.





In most cases the first 3 privileges (TABLE, PROCEDURE and VIEW) will be standard for most schemas that you will set up. So in reality the only command or extra privilege that you will need to execute is:


This command will allow the user to connect to their Oracle schema using ORE, but what it will not allow them to do is to create any embedded R. These are R scripts that are stored in the database and can be called in their R/ORE scripts or by using the SQL API to R (I'll have more blog posts on these soon). To allow the user to create and use embedded R the DBA will also have to grant the following privilege as SYS:


To summarise the DBA will have to grant the following to each schema that wants to use the full power of ORE.



A note of Warning: Be careful what schemas you grant the RQADMIN privilege to. It is a powerful privilege and opens the database to the powerful features of R. So using the typical DBA best practice of granting privileges, the DBA should only grant the RQADMIN privilege to only the people who require it.

Monday, May 12, 2014

Getting Started with ROracle

There are many different ways for you to connect to a database using R. You can setup an RODBC connection, use RJDBC, use Oracle R Enterprise (ORE), etc. But if you are an Oracle user you will want to be able to connect to your Oracle databases and be able to access the data as quickly as possible.

The problem with RODBC and RJDBC connections is that they are really designed to process small amounts of data. As your database and data grows, particularly in the Big Data World then using these type of connections soon become a bottleneck. Another alternative is to use Oracle R Enterprise, but if you do then you have to pay extra licence fees. Again this may not be an option.

An alternative is to use the ROracle package. This R package that is supplied by Oracle, for FREE!!!, allows you to setup connections that utilise the Oracle Client software that you will have installed on your client PCs/laptops etc. Because it utilises the communication technology of Oracle Client you are going to get really good performance and opens up the possibility of processing your Big Data is a reasonable amount of time.

The following steps brings your through the various steps involved in getting ROracle installed, how to connect to the database and how to execute some simple commands. At the end of the post I will point you towards some performance evaluations that have been conducted comparing ROracle to other connection methods.

Installing ROracle (on Client and on the Server)

The first step you need to perform is to install ROracle. If you are installing this on your client machine then you can install it into your R directory. If you are installing ROracle on your server and you have ORE already installed then you can install it in the ORE directory. If you do not have ORE installed on your server but you have R installed then install ROracle in your R directory on the server.

To install ROracle you can run the following command:

install.packages("ROracle") Or select the Install Packages menu option from the R Gui menu or from the RStudio menu. You will get prompted for the R home to install the ROracle package in. ROracle1

You may get some warning messages about some other packages and if the ROracle package was compiled using a slightly different version of R. These are just warning messages and everything should work OK. If you get an error message then you will need to check out what is causing it.

As part of the process the R process will be restarted.

Connecting to your Oracle Database

Now that you have ROracle installed the next step is to test that you can connect to your database. The following commands loads the ROracle package, defines the Oracle driver, sets-up the connection information and then establishes the connection.

> library(ROracle)

> drv <- dbDriver("Oracle")

> # Create the connection string

> host <- "localhost"

> port <- 1521

> sid <- "orcl"

>connect.string <- paste(


>    "(ADDRESS=(PROTOCOL=tcp)(HOST=", host, ")(PORT=", port, "))",

>    "(CONNECT_DATA=(SID=", sid, ")))", sep = "")

con <- dbConnect(drv, username = "dmuser", password = "dmuser",dbname=connect.string)

At this point you are now connected to the DMUSER schema in the Oracle database.

Issuing a Query and processing the results

How we can run our queries on objects in our schema (DMUSER). The following commands sends a query to the schema (to bring back all the view names) and returns the results into rs. The contents of rs is then mapped into data. The dim(data) command returns the number of records in the result set (in the example this is 6) and the number of columns (1 in our example).

> rs <- dbSendQuery(con, "select view_name from user_views")

> # fetch records from the resultSet into a data.frame

> data <- fetch(rs)

> # extract all rows

> dim(data)

[1] 6 1

> data








Checking Query meta-data

ROracle allows us to find out information or meta-data regarding the execution of the query. The following commands list the various meta-data available for the query, then gets the meta-data for the query we ran in the above step (results are in rs) and then displays the meta-data for the rs query results.

> # Get the meta-data about the query

> names(dbGetInfo(rs))

[1] "statement" "isSelect" "rowsAffected" "rowCount" "completed" "prefetch"

[7] "bulk_read" "fields"

> rsInfo <- dbGetInfo(rs)

> rsInfo


[1] "select view_name from user_views"


[1] TRUE


[1] 0


[1] 6


[1] TRUE




[1] 1000


name Sclass type len precision scale nullOK

1 VIEW_NAME character VARCHAR2 30 0 0 FALSE

Checking Database meta-data

You can also get some meta-data about your connection to the database. The following commands lists the meta-data available for a connection, gets the meta-data for the connection and then displays the meta-data for the current connection (con).

> # Get the meta-data about the connection ot the database

> names(dbGetInfo(con))

[1] "username" "dbname" "serverVersion" "serverType" "resTotal"

[6] "resOpen" "prefetch" "bulk_read" "stmt_cache" "results"

> dbInfo <- dbGetInfo(con)

> dbInfo


[1] "dmuser"


[1] "(DESCRIPTION=(ADDRESS=(PROTOCOL=tcp)(HOST=localhost)(PORT=1521))(CONNECT_DATA=(SID=orcl)))"


[1] ""


[1] "Oracle RDBMS"


[1] 1


[1] 1




[1] 1000


[1] 0



Statement:           select view_name from user_views

Rows affected:       0

Row count:           6

Select statement:    TRUE

Statement completed: TRUE

OCI prefetch:        FALSE

Bulk read:           1000

Closing the DB Connection

In this section some commands are given that allows you to see some of the details of the Oracle driver, to Commit any changes to the database, to free up the resources being held by result set of the query we ran in a previous step and then to close the connection to the schema.

> # Free up the resources and disconnect from the database

> summary(drv)

Driver name: Oracle (OCI)

Driver version: 1.1-11

Client version:

Connections processed: 3

Open connections: 2

Interruptible: FALSE

> dbCommit(con)

[1] TRUE

> dbClearResult(rs)

[1] TRUE

> dbDisconnect(con)

[1] TRUE

Performance evaluation

Mark Hornick of Oracle has written a blog post on the performance differences between RODBC, RJDBC and ROracle. He compares the performance of reading from tables with various numbers of attributes and various volumes of records. Check out his blog post here. The following image is taken from this blog post and illustrates the performances.


You can access the R CRAN ROracle Package Documentation here, and the ROracle documentation/webpage here.

Thursday, May 1, 2014

Oracle Magazine-July/August 2000

The headline articles of Oracle Magazine for July/August 2000 were on business intelligence, architectures for BI and how companies like NetFlix m drug-store.com and health insurances companies are using BI to better understand their customers.

2000 July Aug Other articles included:
  • Tom Kyte has an article on Back to Basic for DBAs to ensure robust performance and scalability. He looks at sizing and some of the different aspects involved in this, some of the hot backup methods and the resizing redo log files.
  • Oracle Jdeveloper 3.1 is released and there is a new component of the Oracle XML Developer's Kit (Oracle XDK), which offers end-to-end XML support and provides developers with an integrated, full-featured application development tool for delivering business-to-business applications in Java and XML.
  • Oracle and IBM plan to offer Oracle Internet Directory within AIX and Monterey/64 as a directory option to customers.
  • Over 5 pages of press releases from Oracle Partners.
  • The Common Warehouse Model (CWM) is an open standard by which DWs can share meta-data. Oracle the the OMG are working on defining a standard for sharing metadata.
  • Kevin Loney has an article on transportable tablespaces in Oracle 8i. He discusses issues around generating a transportable tablespace set, plugging in the transportable tablespace set, some management issues and some things to consider with partitions
  • The is a modelling article that looks at the differences between using Relational and an Object-Based model. Taking a real world example the article walks through how to use the relation and then the object model.
  • Richard Niemiec gives the second part of his article on Tuning Problems Queries. In this article he covers using hints, using joins effectively, using the parallel query option and using some mathematical techniques to anticipate query performance.
  • Jonathan Gennick outlines a 6 step incremental approach to developing good SQL queries
  • The final article was an interview with Ari Kaplan who was at the time and Oracle DBA, Oracle book author and talks about his analytics projects with a number of baseball clubs.

To view the cover page and the table of contents click on the image at the top of this post or click here.
My Oracle Magazine Collection can be found here. You will find links to my blog posts on previous editions and a PDF for the very first Oracle Magazine from June 1987.

Tuesday, April 29, 2014

Installing ORE - Part C - Issue installing ORE on Windows Server

In my previous two blog posts (Part-A and Part-B) I detailed 4 steps for how you can install ORE on your servers and on your client machines.

I also mentioned a possible issue you may encounter if you try to install ORE on a Windows server. This blog post will look at this issue and how you can workaround it and get ORE installed.

The problem occurs when I when to install the ORE Supporting packages.

I was prompted to install these into a new library directory. If you get this error message then something is wrong and you should not proceed with installing these packages. If you do proceed and install them in a new library directory then they will not be seen by ORE and the database (as they were not installed in the $ORACLE_HOME/R/library) and when you go to run ORE from within R you will get errors like the following

package ‘Cairo’ successfully unpacked and MD5 sums checked

package ‘DBI’ successfully unpacked and MD5 sums checked

package ‘png’ successfully unpacked and MD5 sums checked

Warning: cannot remove prior installation of package ‘png’

package ‘ROracle’ successfully unpacked and MD5 sums checked

Warning: cannot remove prior installation of package ‘ROracle’

If I try the ore.connect I get the following errors.

ore.connect(user="RQUSER", sid="orcl", host="localhost", password="RQUSER", port=1521, all=TRUE)

Loading required package: ROracle

Error in .ore.oracleQuerySetup() :

ORACLE connection requires ROracle package

In addition: Warning message:

In library(package, lib.loc = lib.loc, character.only = TRUE, logical.return = TRUE, : there is no package called ‘ROracle’

To overcome this ORE install issue all you need to do is to close down your R Gui, then add the following lines to the Rprofile file. The Rprofile file is located in R\etc directory C:\Program Files\R\R-3.0.1\etc. Add the following lines:

# Add $ORACLE_HOME/R/library to .libPaths() for ORE packages


The above line will tell R to look in or to include the R directory in the Oracle home as part of its search path. You many need to change the directory above to point to your Oracle home. When you log into the R Gui the path above will be included. Now you can install the packages and then import the packages. This time they will be installed in the $ORACLE_HOME/R/library.

When you open the R Gui and run the command to load the ORE package and to connect to your ORE schema you should not receive any error messages.

> library(ORE)

> ore.connect(user="RQUSER", sid="orcl", host="localhost", password="RQUSER", port=1521, all=TRUE)

Now you should have ORE installed and working on your Windows server.

Thursday, April 24, 2014

Installing ORE - Part B

This is the second part of a two part blog post on installing ORE.

In reality there are 3 blog posts on installing ORE. The third and next blog post will be on a particular issue you might encounter on a Windows server and how you can over come the issue.

In the previous blog post I outlined the steps needed to install ORE on the database server and on the client machine. Click here to go to this post.

In this blog post I will show you how to setup a schema for ORE and how to get connected to the schema using ORE.

Step 3 : Setting up your Schema to use ORE / Tasks for your DBA

On the server when you unzipped the ORE download, you will find a demo_user.bat script (something similar like demo_user.sh on Linux).

After the script has performed some checks, you will be asked do you want to create a demo schema. Enter yes for this task to be completed and the RQUSER schema will be created in your schema. Then enter the password for the RQUSER.

The RQUSER can as a small set of system privileges that allow it to connect to and perform some functions on the database. This include:





NOTE: If you cannot connect to the database using the RQUSER and the password you set, then you might need to also grant CONNECT and RESOURCE to it too.

For every schema that you want to access using ORE you will need to grant the above to them.

In addition to these grants, if you want a schema to be able to create and drop R scripts in the database then you will need to grant them the addition role of RQADMIN.

sqlplus / AS SYSDBA


NB: You will need to grant RQADMIN to an schema where you want to use the embedded ORE in the database.

Step 4 : Connecting to the Database

If you have complete all of the above steps you are now ready to use ORE to connect to your database. The following is an example of the ore.connect command that you can use. It is assuming the RQUSER has the password RQUSER, and the the host is on the local machine (localhost). Replace localhost with the host name of your database server and also change the SID to that of your database.

ore.connect(user="rquser", sid="orcl", host="localhost", password="rquser", port=1521, all=TRUE);

If you get no errors and you get the R prompt back then you are connected to the RQUSER schema in your database.

To test that the connection was made you can run the following ORE command and then list the tables in the schema.

> ore.is.connected()

[1] TRUE

> ore.ls()


The output of the last line above tells us that we do not have any tables in our RQUSER schema. I will have more blog posts on how you can use ORE and perform various ORE analytics in future posts.

There are a series of demonstrations that come with ORE. To access these type in the following command which will list the available ORE demos.

> demo(package="ORE")

The following command illustrates how you can run the ORE demo called basic.

> demo(basic, package="ORE")

Also check out the Part C blog post on how to resolve a potential install issue on a Windows server.

Tuesday, April 22, 2014

Installing ORE - Part A

This blog post will look at how you can go about installing ORE in your environment.

The install involves a 4 steps. The first step is the install on the Oracle Database server. The second step involves the install on your client machine. The third steps involves creating a schema for ORE. The fourth steps is connecting to the database using ORE.

In this Part A blog post I will cover the first two steps in this process. The other steps will be coved in another blog post.

NB : A the time of writing this blog post ORE 1.4 cannot be installed on a 12c database if it has a CDB/PDB configuration. If you want to use ORE with 12c then you need to do a traditional install that does not create a CDB with a PDB. The ORE team are working hard on this and I'm sure it will be available in the next release (or two or ...) of ORE.

Step 1 : Installing ORE on the Database Server

Before you being looking at ORE you need to ensure that you have the correct version of database. If you have version or then you can go ahead and perform the installation below. But if you have or then you will need to apply a patch to your database. See my note above about 12c.

Download the Oracle R Distribution from their website. Download here.

Although you can use the standard version of R, Oracle R Distribution comes with some highly tuned packages. If you are going to use the standard R download then you will need to ensure that you download the correct version. ORE 1.4 will require R version 3.0.1. Yes this is not the current version of R.

Accept at the defaults during the installation of ROracle, and within a minute or two ROracle will be installed.

Download the Oracle R Enterprise software. Download here. This will include the Server and Supporting downloads.

Uncompress the downloaded ORE files and go to the server directory. Here you will find the install.bat (other other similar name for your platform).

Make sure your ORACLE_HOME and ORACLE_SID environment variables are set.

A number of environment and environment variables are checked. When prompted accept the defaults.

When prompted for the password for the RQSYS user, enter an appropriate password and take careful note of it.

Now go back to the Oracle download page for ORE and download the supporting packages. Unzip the downloaded file. Noting the directory that they were installed in you can now load them in R. To do this open R and run the following commands. You will need to change the directory to where these are located on your server.

install.packages("C:/app/supporting/ROracle_1.1-11.zip", repos=NULL)

install.packages("C:/app/supporting/DBI_0.2-7.zip", repos=NULL)

install.packages("C:/app/supporting/png_0.1-7.zip", repos=NULL)

install.packages("C:/app/supporting/cairo_1.5-5.zip", repos=NULL)

Or you can use the R Gui to import these packages

WARNING:If you are installing on a Windows server you may encounter some issues when importing these packages. I will have a separate blog post on this soon.

NB: The ORE installation instructions make reference to Cario-_1.5-2.zip. This is incorrect. ORE 1.4 comes with Cario-_1.5-5.zip.

At this point, assuming you didn't have any errors, you now have ORE installed on your server.

Step 2 : Installing ORE on the Client

Download the Oracle R Distribution from their website. Download here.

NOTE: If your database and client are on the one machine then there is no need to install ROracle again.

The client install is much simpler and less involved. After you have installed ROracle the next step is to install the client packages for ORE. These can be downloaded from here.

After you have unzipped the file you can use the import packages from zip feature of the R Gui tool or using RStudio. Then import the supporting packages that you also installed as part of the server install.

Now you can install the supporting packages. Unzip them and then use the R Gui or RStudio to importing them. These supporting packages can be downloaded from here.

That should be the client R software and ORE packages installed on your client machine. The next steps is to test a connection to your Oracle database using ORE. Before you can do that you will need to setup a Schema in the database to use R and also grant the necessary privileges to your other schemas that you want to access using R

Check out my next blog post (Installing ORE - Part B) for Steps 3 and 4.

Also check out the Part C blog post on how to resolve a potential install issue on a Windows server.

Monday, April 14, 2014

Oracle R Enterprise and Oracle 12c

A few of weeks ago we had the release of Oracle R Enterprise (ORE).

There has been some posts on the R/ORE on the Oracle discussion forums about installing ORE on Oracle 12c.

It turns out that the only way to install ORE on an Oracle 12c database is if you do a traditional install. What this means is that you do not have a CDB and PDBs configuration of Oracle 12c.

I'll assume that Oracle are currently working on this particular issue, as you can imagine that that there is considerable amount of complexity in getting ORE to work with the PDBs.

If you are not using Oracle 12c then you are OK, as long as you are using or versions of the database. If you are using a lower version of the 11.2 database then you need to apply a patch to allow ORE to run.

As they say I'm sure it will be "fixed in the next release" :-)

Oracle Advanced Analytics and Oracle Fusion Apps

At a recent Oracle User Group conference, I was part of a round table discussion on Apps and BI. Unfortunately most of the questions were focused on Apps and the new Fusion Applications from Oracle. I mentioned that there was data mining functionality (using the Oracle Advanced Analytics Option) built into the Fusion Apps, it seems to come as a surprise to the Apps people. They were not aware of this built in functionality and capabilities. Well Oracle Data Mining and Oracle Advanced Analytics has been built into the following Oracle Fusion Applications.
  • Oracle Fusion HCM Workforce Predictions
  • Oracle Fusion CRM Sales Prediction Engine
  • Oracle Spend Classification
  • Oracle Sales Prospector
  • Oracle Adaptive Access Manager
Oracle Data Mining and Oracle Advanced Applications are also being used in the following applications:
  • Oracle Airline Data Model
  • Oracle Communications Data Model
  • Oracle Retail Data Model
  • Oracle Security Governor for Healthcare
I intend to submit a presentation on this topic to future Oracle User Group conferences as a way of spreading the Advanced Analytics message within the Oracle user community. If you would like me to present on this topic at your conference or SIG drop me an email and we can make the necessary arrangement :-)

Friday, April 11, 2014

Oracle 11.2g install on OLE 6.x

This notes are really just a reminder to myself of the typical "issues" that I encounter every time I do a new install of OEL 6.x and

These notes are in addition to the excellent installation instructions given by oracle-base.com: oel install, DB 11.2.0.x install

The notes listed below are just a reminder to myself of things that I seem to always have to look up. If you finish them useful then great.

1. Display issue & Installer not able to run

install says to do xhost +:0.0 this can give an error

instead do host +:0.0 and that should allow the installer to run

2. Now enough swap space when installer checks the pre-requisites

Need to add an addition 500M to the swap space

su (and then enter the password)

dd if=/dev/zero of=/tmp/swapfile bs=1M count=500

mkswap /tmp/swapfile

swapon /tmp/swapfile

exit (to return to the oracle user)

you can then turn off the extra space (if you really need to) after the install is finished

swapoff /tmp/swapfile

rm /tmp/swapfile

3. Post-Installation task

don't forget the final step, to set to restart flag

as root

vi /etc/oratab

change the following line to have the Y at the end (instead of the N)


4. Set up the automated start and stop of the DB

Again Oracle-Base gives an excellent set of instructions for doing this. Click here.

Wednesday, April 9, 2014

Oracle Text and Oracle Data Miner

This blog post is a follow up to comment on a previous blog post and to some emails.

Basically the people are asking about some messages they get when they open the Oracle Data Miner tool, that is part of SQL Developer.

If you are just using the SQL and PL/SQL functions in the database then you do not have to worried about Oracle Text. You will receive no warning message.

But if you use the Oracle Data Miner tool you will get a warning message.

Why do you get this message? Some of the functionality in the Oracle Data Miner tool relies on having Oracle Text enabled/installed in the database. You can locate this functionality under the Text section of the Component Workflow Editor palette of Oracle Data Miner.

So if you are getting these warning messages then Oracle Text was not installed when the database was created.

How can you install Oracle Text? There are 2 scripts that you need to run.

For the first script you will need to log into SYS as SYSDBA and run the following script.

ctx/admin/catctx.sql password SYSAUX TEMP NOLOCK

This script will create a user called CTXSYS with the password of password (give above), with the default tablespace of SYSAUX, the temporary tablespace of TEMP and when the account is created don't lock it (NOLOCK).

This script will also install a number of CTX packages.

The next step is to log into the CTXSYS schema (using the password above) and run the following script.


This takes a parameter to specify the language you want to use. For example "English", "AMERICAN", etc.

The final step is to connect as SYS again and lock the CTXSYS account.

alter user ctxsys account lock password expire;

If you are using Oracle 12c then the above steps will be automatically done for you during the process. If you are using an earlier version of the database or a database that has been upgraded through some version then Oracle Text may not have been installed. In this case you can run the able commands.