Bloggers physically present at OpenWorld 2009 list


Updated with latest confirmed attendees of the Bloggers Meetup tonight. Marcel Kratochvil, Jacco Landlust, William Vambenepe, Stuart Marks, Avrom I. Roy-Faderman, Stanley ACE Director, OCP Advisor

There are always more to add, so please check back (and give feedback; I’ve received most via Twitter – eespenh – and that’s great). Since last, I’ve added Daniel Morgan, Karen Morton (because I noticed them standing in the Unconference area), Fenng, Khanderao, Merv Adrian.

Since yesterday here at OpenWorld, I noticed that Bex Huff, Hans Forbrich, Moans Nogood (not really) and Jason Jones are also here. And Claudia Zeiler (girlgeek), whom I just met outside the OTN lounge. Wonder if I’ve got half of everyone before the week’s over? – Avrom I. Roy-Faderman – Ted Simpson – Bex Huff – Arun Gupta – Arjen Visser – Raimonds Simanovskis – Chris Bucchere – Cary Millsap – Todd Sheetz – Dietmar Aust – Fenng – Debra Lilley – Dimitri Gielis – David Peake – OCP Advisor – Markus Eisele – John E. Bredehoft – Marcel Kratochvil – Fuad Arshad – George J. Trujillo, Jr. – Espen Barroso-Gomez – myself – Gareth Roberts – Claudia Zeiler – George Woods – Hans Forbrich – Jason Arneil – Kohsuke Kawaguchi – Jonathan Lewis – Karen Morton – Khanderao – Matt Topper – Merv Adrian – Michelle Malcher – Dave B – Tim Hall – Jonah H. Harris – Doug Burns – Jacco Landlust – Jeff Browning – Floyd Teter – Gwen Shapira – Jason Jones – Andrew Clarke – Richard Foote – Roel Hartman – Rob Van Wijk – William Vambenepe – Stuart Marks – Lutz Hartmann – Tim Bray – Lucas Jellema – Tom Kyte – H. Tonguç Yilmaz – Moans Nogood – Marco Gralike – Daniel Morgan – Douwe Pieter van den Bos – Alex Gorbachev – Stanley ACE Director

Others who might be here or have more than one main contributor: – R “Ray” Wang – Mark Bennett, Meg Bear, Amy Wilson

OpenWorld 2009 – who’s there?


I’ll be going over there myself in a few hours, crossing the nine hour time difference. But I started wondering which of my fellow, admittedly much more industrious, blog writers, would be present at Oracle OpenWorld 2009 (#oow09) as well. I’ve read their notices, but my memory isn’t good enough to piece that into a picture.

So here’s a first attempt to list those blogs whose author has expressed an intention to attend OpenWorld this year. This is just the result of one hour of searching through and Google, and not counting blogs under Some blogs I’ve left out because it’s unclear to me which one person represents it best; please leave a comment in cases where I’ve gotten things wrong, though! As an example, isn’t really Judi Doolittles personal blog, even if she’s responsible for the content, but she might have wanted it in such a list anyway?

So I’m just making up the rules as I go, I guess (why? because it’s two o’clock in the morning (the taxi for the airport arrives in two hours) :)
I’m probably missing a lot of persons in this first attempt, anyway.

The number of people attending OOW according to the event (created by Ontario Emperor) in Oracle Community ( at present is 7. So if you’re a member of Oracle Community, do sign up for the event, and if not, do sign up for Oracle Community. And Oracle Mix ;) (But Oracle Mix won’t let me in this week; that’s another story, I guess.)

Another digression:
The number of bloggers attending the Blogger Meetup event (which is Tuesday at 6pm at Jillians, by the way; look it up, is around three times as many. Looking forward to meeting many of you there.

So this is the list at the moment: – Avrom I. Roy-Faderman – Ted Simpson – Bex Huff – Arun Gupta – Arjen Visser – Raimonds Simanovskis – Chris Bucchere – Cary Millsap – Todd Sheetz – Dietmar Aust – Debra Lilley – Fenng – Dimitri Gielis – David Peake – OCP Advisor – Markus Eisele – John E. Bredehoft – Marcel Kratochvil – Fuad Arshad – George J. Trujillo, Jr. – Espen Barroso-Gomez – myself – Gareth Roberts – Claudia Zeiler – George Woods – Hans Forbrich – Jason Arneil – Kohsuke Kawaguchi – Jonathan Lewis – Karen Morton – Khanderao – Matt Topper – Merv Adrian – Michelle Malcher – Dave B – Tim Hall – Jonah H. Harris – Doug Burns – Jacco Landlust – Jeff Browning – Floyd Teter – Gwen Shapira – Jason Jones – Andrew Clarke – Richard Foote – Roel Hartman – Rob Van Wijk – William Vambenepe – Stuart Marks – Lutz Hartmann – Tim Bray – Lucas Jellema – Tom Kyte – H. Tonguç Yilmaz – Moans Nogood – Marco Gralike – Douwe Pieter van den Bos – Daniel Morgan – Alex Gorbachev – Stanley ACE Director

Others who might be here or have more than one main contributor: – R “Ray” Wang – IOUG / Judi Doolittle, Steve Lemme – Sharat Chander, JB – Shyam Varan Nath, John Haydu – Martin Nash, Neil Johnson, Pawel Krol – Mark Bennett, Meg Bear, Amy Wilson

Installing 11gR2 on OEL 5.3 – postrequisites


A quick followup on my post on installing 11gR2 a little while ago (last post).

There were actually only two things I had problems with after this Oracle install on OEL 5.3:

  1. I couldn’t start the listener (got an error like “cannot restore segment prot after reloc: Permission denied”) until I had set SELinux to Permissive:
    setenforce 0
    (See for example for a discussion on this)
  2. I couldn’t reach the administrative URL’s before I had configured/disabled the local firewall, for this test instance with the quick fix:
    service iptables stop

Installing 11gR2 on OEL 5.3


These are notes and first impressions of the install process of 11gR2. All my comparisons are with the 10g installer, since I realized during this install that I haven’t ever tried the 11gR1 installer (I’ve set up 11gR1 databases and instances, but using templates and other stuff, not the installer). So bear in mind that some of the details I noted, might not be as new as I think.

The setting:

We have a Dell 2850 which I’ve recently reinstalled with a plain (64-bit) Oracle Enterprise Linux 5 update 3 (most recent one at this time). Since the availability of the new database release coincides with this currently idle database server, it’s a perfect time to see how the new release behaves.

This will be a generic install with no specific requirements from my side, and if everything works as expected, I’ll try to see how an import of a 10gR2 database into it works.

I haven’t prepared the OS in any way (except one, see later), since I also want to see how the install handles the prerequisite checks and rechecks (in previous releases I’ve often had to restart the installer for it to pick up certain changes).

The one thing that has been done to the OS after install is adding the Oracle Public Yum server repository to /etc/yum.repos.d/ in case I need any additional packages later (the public-yum-el5.repo file from now has a separate el5_u3_base entry).

Now on to running the installer.

It’s looks like a 9 step installer, with an overview of all the steps before any questions are asked. Nice touch. But then, it appears that it’s actually a 9 step installer only in certain combination of choices. The number of steps will change according to the choices made.

The starting steps are:

  1. Configure Security Updates (email/login for My Oracle Support)
  2. Installation Options (create databases, install software only, upgrade)
    2.5. System Class (desktop, server)
  3. Grid Options (single instance, RAC)
  4. Install Type (typical, advanced (different account passwords, char. sets, ASM and so on))
  5. Typical Installation (installation details like paths and database name)
  6. Prerequisite Checks
  7. Summary
  8. Install Product
  9. Finish

I chose the advanced installation at this time, which caused step 5 to split up into separate (and familiar) steps: language selection, database edition, installation location, oraInventory/os group, type of database, global name/sid, configuration (memory, character sets, opt out/in of new security settings, sample schemas), OEM Grid Control or not, file system or ASM storage, enabling backups, passwords, os groups)

And then comes the prerequisite checks. Which failed in three areas.

Here’s my missing dependencies, on top of a plain OEL 5.3 install:

  • Editing /etc/security/limits.conf; adding oracle hard nofile 65536
  • Editing kernel parameters in /etc/sysctl.conf, setting
    • kernel.sem = 250 32000 100 128
    • fs.file-max = 6815744
    • net.ipv4.ip_local_port_range = 9000 65500
    • net.core.rmem_default=262144
    • net.core.rmem_max=4194304
    • net.core.wmem_default=262144
    • net.core.wmem_max=1048576
    • fs.aio-max-nr=1048576
  • And the following packages were missing
    • libaio-devel
    • sysstat
    • unixODBC
    • unixODBC-devel

The missing packages were listed twice, for some reason. Perhaps once for the 32-bits and once for the 64-bits version.

But now the installer provides a tailored script to fix these thing right away, which I haven’t seen before, the script. Unfortunately I went ahead and fixed these things manually before noticing the script, so I haven’t tried it yet. Looking at the code (in /tmp/CVU_11., it seems it would easily fix the limit and kernel settings, and it’s supposed to install the missing packages as well, by using YUM. But when I looked at the fixup.enable file, it only flagged the former to things to be fixed:


So I have to assume some extra steps would have had to be taken for the software package download/install to work. Do any of you have experience with this?

Anyway, in my case, I simply installed the repositories as mentioned above, and the packages, like this:

cd /etc/yum.repos.d/
vi public-yum-el5.repo (to set enabled=1 for el5_u3_base)
yum list
yum install libaio-devel
yum install sysstat
yum install unixODBC
yum install unixODBC-devel

And that was more or less it. I just saved the response file for another day, and started the install, which finished in around 10 minutes.

One thing to note, is to make a copy of the emkey.ora file. I suppose (or I want to suppose) that most people backup their ORACLE_HOME directories properly, so for disk loss situations, you’d probably restore all files therein, but it does hurt to make the copy in case of accidental deletions.

After the install finished, there are the usual “run as root” scripts to execute.

Additionally, there are a few things to fix before it works, because of settings in the default OEL installation (which I believe Oracle should have been able to fix before this), but I’ll get back to these in another post in a few days (or if anyone asks in the meantime :)

Using ASM with ASMlib or raw devices


Another post to fix some things better in memory.

Some typical tasks when using ASM, possibly with Oracle VM

If virtual, create the disk first
Either through the Oracle VM Manager interface (Virtual Machine Configure -> Storage -> Create New Virtual Disk), or by stopping the virtual machine and editing the vm.cfg file (adding lines to the disk clause like for example: ‘file:/OVS/running_pool/21_db01/asm01.img,xvde,w’,). In the latter case, these new virtual disks must exist before the vm is started again, typically by creating zero-filled files with dd (place in the directory specified in the vm.cfg):
dd if=/dev/zero of=asm01.img bs=1M count=50000

I don’t know of a way to make the virtual machine pick up changes like new disks in the vm.cfg without restarting, but there should be one, since doing the same thing through the OVM Manager interface does not cause any rebooting. I’ll make a note to figure this out, but any feedback from someone who might know is appreciated.

Partition each storage disk with one primary partition per disk, using fdisk.
fdisk /dev/sdX

Configuring Oracle Cluster Synchronization Services (CSS)

Run localconfig as root (which probably also means you need to exchange the value for the ORACLE_HOME manually):
$ORACLE_HOME/bin/localconfig all

Setup for ASMlib
Installation: If not already present in OS; download oracleasm packages from OTN, install and configure (also as root)
/etc/init.d/oracleasm configure
Stamp disks for use with ASMlib:
/etc/init.d/oracleasm createdisk VOLn /dev/sdX1
scandisks if using with additional RAC nodes, or listdisks for verification.
/etc/init.d/oracleasm listdisks

Setup for raw devices
Edit mapping file /etc/sysconfig/rawdevices (rawpath devicepath)
Restart rawdevices service
service rawdevices restart
Set ownership (oracle:oinstall) and mode 0600 for newly defined raw devices, and add to /etc/rc.local as well.

ASM instance creation
Use DBCA, “Configure Automatic Storage Management”, follow the instructions (about CSS and passwords), create new disk group.

Database creation
Verify the listener is running and that +ASM is registered. If not, register with alter system register, and check lsnrctl status again.
Choose ASM when creating the database, using the ASM disk group (typically +DATA) for Oracle-managed files, flash recovery area and archiving.

Switching  from raw/ASMlib and back

  1. Shutdown other instances, connect to +ASM.
    Install ASMlib, without the disk stamping.
    Set ASM_DISKSTRING to the preferred label for the disks, ex. ‘ORCL:VOL*’, shutdown +ASM.
    Stamp the disks, but use renamedisk, since createdisk would fail when the disks have already been used raw.
    /etc/init.d/oracleasm renamedisk /dev/sdX1 VOL1
    and so on.
    Startup +ASM and other instances.
  2. Shutdown other instances, connect to +ASM.
    Set ASM_DISKSTRING to wherever the raw devices can be found; ex. ‘/dev/raw/raw*’, shutdown +ASM.
    Perform raw device setup (edit mapping, start service, set file ownership).
    Startup +ASM and other instances.

Enabling Automatic Archiving


Just a post to remind myself how to start archiving. There are a few details I tend to forget, and presumably they stick better to memory when written down.

Fact 1: ALTER SYSTEM SET LOG_ARCHIVE_START=TRUE SCOPE..(static variable, remember bouncing to effectuate)
Subfact: Definition of “archiving manually” is to start the arch process without setting the init parameter. (ALTER SYSTEM ARCHIVE LOG START)

Status of my hiatus


Should have been the name of my blog for a long time.

It’s been a while. It’s also been a while since I got back at work (after a period of  paternity leave following Openworld).

I’ve got to say I was a little naive on how many blogging hours I’d get around to putting in during this (paternity leave) time. (End result: None). But it was  a fantastic and irreplacable slice of life, so I can hardly regret the priorities. I did get time to tinker a bit with Oracle VM, though.

The rest of the time, roughly since New Year, I’ve wrestled with a different problem: Normally, I would have had several technology projects ongoing, but for the time being, I’ve been needing to focus exclusively on marketing and sales related activities. Which isn’t really what I began this blog to write about, and since I’m a bit new to that, it’s unclear to me what insights I could contribute here.

What will be, will be, though. I have been working on a few things, and will make a second attempt to start up this channel :) Incidentally, I was encouraged to see Bex Huff, whose blog is one of the most useful I know of, writing that his blog is actually his fourth one:

There might be hope for me, as well.

So the status of my hiatus is that it’s hopefully over for now.

Fusion – are we there yet?

Fusion Apps, Middleware, Oracle, Uncategorized

Fusion – are we there yet?

I’ve been explaining Oracle Fusion many times. But I’m starting to feel I’m not getting somewhere
fast enough..

It’s always been a little fuzzy what Oracle Fusion entails, since the first announcement in 2005.
(I have no memory of Fusion from before that time, so first mention might have been even earlier.)

Myself, I’m a big fan :) since all that Fusion means, generates potentially a lot of of work for
people like me, albeit possibly a bit into the future still, at least in Norway, and definitely in these
murky times. Even more, I’m a fan because it gives me more technology to play with under the
Oracle umbrella, leading to more novel ways of doing integration.

I attended at least one session at Oracle OpenWorld this year that aimed to explain what
Oracle Fusion is, and I’m still less than impressed, by the fact that there’s still so great a need for
explanations. (But that session was well done; S301194)

Not at first; at the time I was just soaking up perspectives and information, but after a few more
days in San Francisco discussing Fusion with people, I was a little disappointed:

– Oracle launched Fusion more than three years ago as their most important strategy, campaign
and brand, but even after this time, I hear customers talking about conFusion. Isn’t it time to get
this sorted out?

– As can be seen from, Fusion is a lot of stuff (“stuff” has a woolly sound to it, which fits
well in this context..). That doesn’t have to be bad, but I really would have liked a more streamlined
presentation from Oracle. At the present, the structure (for all things Fusion) is like a tree, where
you could start at any branch, and try to work your way downwards to the trunk. What I’m trying to
say is: I can’t give a client a/my description of Fusion, and then point them to for more
information; I’d have most of them lost in translation.. (I admit I wrote this from my general feeling
of things, so I had to give it a try. Going for example to will lead you to the Fusion
Applications sales page. Most of it assumes you visit because you know what Fusion Applications meant
to replace. But I have to qualify my statement; there’s a “Have Oracle call you” button, so if you’re
curious enough, you might click that and you might get the Fusion introduction you’re looking for.
But I haven’t tried, and I seriously doubt it :)

– A lot of money goes into the Fusion Applications product line (I must assume), and undeniably a lot
goes into Fusion Middleware (see BEA, etc). Most goes into development and acquisitions. Could it be
that marketing has a hard time keeping up?

To wrap it up: We definitely need a better definition of Fusion.

The integration of all things BEA into the products and brand should be an excellent opportunity to make this a priority.

Abstracts for OUGN


I’ve submitted two abstracts for the Norwegian Oracle User Group ( forum in March 2009, one for E-Business Application Management Pack and one for Oracle SOA HA (Fusion Middleware Grid) testing.

I think the abstracts are a little plain and simple, but as this was my first attempt, and given that the actual presentations haven’t been fully finished yet, I hope it’s a good start.

Now I’ll just have to wait a few weeks to hear if they are accepted.

Abstract for SOA HA:

This presentation will describe how to set up a test Enterprise Deployment architecture of the Oracle SOA suite, by showing how to deploy a Fusion Middleware Grid environment with two or more machines, in an active-active state, based on either Oracle WebLogic Server and Oracle Application Server.
It will feature an introduction to virtualized environments, and will demonstrate how Oracle VM templates can be used to greatly speed up the process of installing, patching  and configuring the machine nodes for working together in a grid.
Also, an overview of the high availability capabilities and technologies WebLogic Server and Oracle Application Server uses to interact with other nodes in a grid, and how to configure these, as well as issues to consider if using Oracle RAC in the underlying database.

Abstract for EBS AMP:

This presentation will discuss the key features and possible benefits of using the Application Management Pack (AMP) for managing the Oracle E-Business Suite, shown by use cases like patching, cloning and monitoring. The presentation will describe the differences of using AMP with an R12 system as opposed to an 11i system, and will discuss the enhancements that have been added to the most recent version of the tool on the way towards enabling true one-click hot-cloning of EBS systems.
Other items discussed include a comparison with similar features delivered by other Oracle utilities, such as the Data Masking feature of Oracle Enterprise Manager vs the AMP Data Scrambling, and comparison in general between AMP and Oracle Applications Manager.

What’s the point of OpenWorld


There a lot to see, and even something new the learn once in a while here at OpenWorld :) But those are the simplest reasons to go here, apart from networking with peers, and they are not the easiest to sell to management (if you are a technical person, I mean; for sales people it appears to be an easier sell).

I’ve had this discussion with several Norwegian companies that does not send anyone over, at least no DBAs or architects. And of course they look at cost as the decisive matter for this, saying that it’s a better return on investment to attend local user group and Oracle University classes.

It’s not.

The user groups are immensely useful, but there’s just no comparison with the added dimension of raw inspiration to be had at OpenWorld, if you go with the flow and use the atmosphere here to get creative.

The real value of shipping employees over to San Francisco once in a while is not easy to put a coin value to, but I really think is it underestimated by many; it is having them return motivated (which is often good for loyalty), and potentially full of new ideas or ways to view their everyday problems and tasks (which is always good for productivity).
It might not be quite priceless, but it shouldn’t be ignored. (I was about to say “can’t be”, but obviously that’s just what’s happening :)

What are your experiences with this? Have anyone else met this reluctance from management for the same reasons?