Coopers Consulting Solutions Limited

Looking at the notes i’d written when i did this i thought it best to break this up into 2 parts.

This first part looks at the installation and configuration of the server. Part 2 will focus on the install of R12 and the post install steps.

Right-o let me set the scene. This install is onto a 1-node-1-user machine; i.e. a laptop
I’ve done the install before, and that was onto a laptop with Oracle Enterprise Linux (OEL) 4.7. However this time i wanted to step away from the usual enterprise distro’s and go with one of the more common respin’s.

For the uninitiated, CentOS is based on Red Hat enterprise – So 5.2 (or 5.3 in this case) is built on RHEL 5.2

Hardware

For this install, i am using a Dell Vostro 1710 laptop that i bought for the sole purpose of putting Linux and my linux development toolkit onto. The core specs are as follows:

• CPU: Intel Core 2 Duo T5870 (2.0 GHz, 800MHz FSB, 2MB cache)
• Display: 17in widescreen WUXGA (running an impressive 1900×1200 resolution)
• Graphics: The usual Dell fare of NVIDIA GeForce 8600M GS (featuring 256MB DDR2)
• Memory: 4Gig – 667MHz DDR2 Dual channel
• Hard Drive: 500GB (Dual 250GB Seagate Momentus G-Force 7200rpm 16MB cache SATA) – This was not standard on the Dell configurator so was a post deliver mod
• Network:  Fairly standard Realtek RTL8168C PCI-E Gigabit ethernet, which fits nicely with my wired Gigabit network at home
• Optical Drive: 8x DVD+/-RW/RAM Drive

It came shipped from Dell with a downgrade to XP. The XP install lasted long enough for me to boot it to verify that all hardware was present and correct before the whole lot was wiped.

OS Install

The media for the install was a free DVD supplied with the December edition of linux format magazine. I’d already tried it out under Virtual Box, running on my OEL4.7 install on the same hardware, and had been impressed that it had nailed the Soundcard and Wireless card straight away (unlike the OEL install).

Anyway, booting up from the DVD i went through the partioning configurator and set up the following:

• The dual HDD’s were configured as a single logical volume (500GB)
• The boot partition /boot was configured to be 15MB fixed size
• The swap partition /swap was configured to be 4096MB (ie same size as the installed memory)
• The root partition / was configured to fill the remaining available disk space. I had to double check this, as my first attempt using automatic partitioning did not fill the remaining space and i was left with a large area effectively closed off from the OS!

Important!
Make sure that you elect to have NO firewall, and that the “Enable SELinux?” option is set to DISABLED. This is advice given in the original notes i followed when doing the OEL install last year. Aparently it causes issues later on when you come to install EBS R12.

Package Groups

During the OS install you will face the package manager. Presenting you with an array of scintillating choices. As i had disk space to play with i went a bit mad and selected the following:

• X Window System (required for EBS)
• GNOME Desktop environment (also required for EBS)
• Editors (your choice – i’m a big fan of Vi )
• Graphical internet
• Server configuration tools
• Development tools
• Administration tools

• System Tools

Required Packages

This list is taken from the Metalink note 402310.1 – Oracle Applications Installation and Upgrade Notes Release 12 (12.0.4) for Linux 32-bit

The following are required packages for EBS R12 (running on OEL 5 but also for CentOS). All versions are minimum required versions (except where stated):

• compat-glibc-2.3.4-2.26
• gcc-4.1.2-14.el5
• gcc-c++-4.1.2-14.el5
• glibc-2.5-12
• glibc-common-2.5-12
• glibc-devel-2.5-12
• libgcc-4.1.2-14.el5
• libstdc++-devel-4.1.2-14.el5
• libstdc++-4.1.2-14.el5
• make-3.81-1.1
• gdbm-1.8.0-26.2.1
• libXp-1.0.0-8.1.el5
• libaio-0.3.106-3.2
• libgomp-4.1.2-14.el5
• sysstat-7.0.0-3.el5
• compat-libstdc++-.296-2.96-138
• compat-libstdc++-33-3.2.3-61

In addition, i needed to pull the following files from the Oracle compatability project archive, which is found at http://oss.oracle.com/projects/compat-oracle/files/Enterprise_Linux

• openmotif21-2.1.30-11.EL5.i386 (this must be this version exactly!)
• xorg-x11-libs-compat-6.8.2-1.EL.33.0.1.i386
• binutils-2.17.50.0.6-6.0.1.i386

In checking whether you have all the required packages, you can query each package individually using:
rpm -q glibc
etc

or you could query all packages and put the output into a text file, and then view this file to search for each package:
rpm -qa >rpm_files.txt

view rpm_files.txt

For the files downloaded from Oracle, i placed then into a separate directory and them manually installed them using rpm:
rpm -Uvh libaio-0*

Software requirements

The following maintenance tools must also be installed and their locations specified in the PATH of the account that will run Rapid Install. Both times i’ve done this install these programs have (a) been installed, and (b) present in the path.

• ar
• gcc
• g++
• ld
• ksh
• make
• X Display Server

You can easily verify these by using the which command:
which make
>> /usr/bin/make

My final step to finish off the OS install itself was to run the automatic updater yum update as this applied a load of patches to take the OS to 5.3.

Onto the home straight now for the server config…

Kernel Settings

I then had to edit the file /etc/sysctl.conf to configure my kernel settings. The following table is taken from the metalink note i referred to earlier.

kernel settings for EBS R12 on Linux

DNS resolver parameters

Two Domain Name System (DNS) resolver parameters are set by default to low values when the OS is installed. These low values may cause attempted network connections to an Oracle database to fail. It is reccomended that these minimum settings are set in the /etc/resolv.conf file:
options attempts:5
options timeout:15

Verifying Host Names

I verified that the /etc/hosts file was formatted as follows:
127.0.0.1 localhost.localdomain localhost
<ip_address> <node_name>.<domain_name> <node_name>

I then had to verify that the /etc/sysconfig/network file was formatted as follows:
HOSTNAME=<node_name>.<domain_name>

If the /etc/sysconfig/networking/profiles/default/network file exists then delete it! No such file existed on my install so i skipped this step.

I then restarted my laptop (again).!

Other settings

I modified the /etc/security/limits.conf file to ensure the values for “hard” and “soft” open file descriptor limits were at least the minimum as given below:

* hard nofile 65536
* soft nofile 4096
* hard nproc 16384
* soft nproc 2047

It states in the install notes (metalink) that if any of these values are already higher then not to lower them!

Oracle HTTP Server Patch

This step was originally missed out when i first tried this install. Shame on me, as it caused no end of troubles. It was only when i looked back at the metalink notes that i realised i’d missed it and it was rather important!

I needed to download and install the patch 6078836 from Oracle Metalink. It fixes an issue with the Oracle HTTP server that is bundled with EBS. Its only applicable to OEL5 or RHEL5, but as we’ve established, CentOS is a RHEL respin so needed to apply as well. For those wondering….I have a Metalink account so was able to do this. In fact i actually have 2 accounts; i have one linked to the site that i’m currently contracted to, and i also have a personal account linked to an OEL license pack that i purchased last year. Probably the best $100 my company has spent in ages

Once i’d studiously followed the patch install instructions, i then had to perform the following task; essentially this is a patch post-install step to link the Motif library for OAS 10.1.2 – this needed to be done as root:

unlink /usr/lib/libXtst.so.6
ln -s /usr/X116R6/lib/libXtst.so.6.1 /usr/lib/libXtst.so.6

And there you have it. End of part one. We have a configured server waiting for an install of EBS

Crikey! Today’s news is a bit of a shocker!

Admittedly Sun have been on shaky ground now for some time, but Oracle! Where did that one come from!

I was never an employee of Sun, but did have the good fortune to work on an excellent project for them during my time at Oracle. They were implementing the Oracle HRMS suite, and the project was being run out of their offices in Palo Alto, and was called GEMS (Global Employee Management System if my memory serves me well). This would have been circa 98/99 and early 2000 – just prior to the Dot-Com bubble bursting I guess.

As the main Apps Technician from Oracle (their budget stretched to one Apps Technical Expert, one functional expert, a support rep and an occasional DBA flown in for good measure), I got to work with all of the Sun technical teams involved with the project, be they reports, forms, interfaces or data migration teams. When not on-site in Palo Alto I worked from a hot desk based out of their UK offices in Camberley (and latterly out of Fleet), and occasionally from my dining room at home (when I needed to do US time in the UK!).

This meant a fair deal of shuttling between UK/US and thus I managed to accrue a good collection of frequent flyer miles

I remember my first day on the project… having just stepped off a 14 hour flight from Heathrow to San Francisco and a quiz sprint down the 101 to their offices, and after all-to-brief introductions i was whisked into what can only be described as an interrogation. Me on one side of the desk, a mass of Sun people on the other, in an office next to the server room (and so no windows, no clocks). The purpose was to “quiz” me to see if I was any good and the right person for the project.

I can’t recall everything that happened, but I do recall that I had technical questions fired at me in a seemingly random fashion from all and sundry, with questions not necessarily being in order…So I’d field an Oracle forms question, then a detailed one about Employee API’s, followed by a reports question and then back to a forms question….This seemed to go on for hours and hours, but in reality it was probably only 60 minutes – 90 at a push.

I must have given a good account of my self for the following morning I was presented with a filled agenda for my 2 week stay, effectively being shuttled between the various teams in a bid to get them all rolling.

In all I think I had 18 months of involvement in the project, and looking back I still regard my time on the project as being one of the most enjoyable periods in my (almost) 12 year tenure at Oracle.

Anyway, the reason for this trip down memory lane is partly because of the hardware. Sun Sparc workstations filled every available crevice – some of the engineers I worked with had several, some of which seemed simply to be dumped on the floor as newer kit had arrived to replace….
It was my first real exposure to a desktop OS other than Windows since my early days of employment (when I used Data General kit featuring their flavour of Unix). The kit was good, robust, effective in use and (get this) did not need constantly rebooting I think I was hooked. It certainly got me started on the Linux bandwagon that’s for sure.

I also recall something called a “Sun Ray”. It was a thin client machine, with a smartcard reader built into it. You plugged in your card and up popped your desktop, complete with apps and anything you happened to have been working on. You could take out the smartcard, wander over to another machine, plug in and be working where you left off immediately. The rumour from the Sun guys I worked with was that you could plug your card into any Sun Ray, even if it were located in a different office in another country, and be up and running instantaneously. Whether this was just rumour of truth I’ll never know, but what does amaze me is why this technology never took off?

Anyway. Sun are being swallowed up by the corporate behemoth that is Oracle, and thus another piece of history will vanish. I wont even start on the speculation as to what is going to happen to MySQL? I’d like to think that Oracle see past the balance sheet and realise that we still need a truly free Open Source database, and leave it well alone (maybe someone out there would like to buy it off them?).

Until then I shall keep half an eye on the news coming out of Redwood Shores to see what their plans are.

A dangerous thing the internet is…

Working in IT as i do (and as i have done for all of my working life) i have often leapt with open arms onto new technology. Often this took the form of buying into gadgetry, often at great expense, and often only for a short-lived buzz of excitement… Who remembers the Iomega Zip Drive? Or the Creative DAP Digital Music Player, or even the Philips Pronto (or Marantz RC7000) all-in-one programmable remote control? I do…I did buy these things (more or less as they came out)…I did use these things, even when they were not quite fit for purpose (the Iomega was restricted to its proprietry media cartridge system, the DAP had a mere 6gig of storage, and the Pronto was ludicrously difficult to program)…

Other times i have been slow to adopt (or conform as many of the younger generatation would see it). Having burnt my fingers on the Creative DAP, i studiously avoided joining the iPod generation until they brought out a unit with (what i considered to be) a sensible amount of storage. When they did bring out such  version i had no further cause to delay (i am the proud owner of a 160gig iPod and have it chock-full of music – no photo’s just music).

Anyway, this brings me to the general gist of today’s rambling… Web 2.0… Specifically the cult of revealing ones life – full story and pics – to the general masses of the internet public.

I’m happy to say that, until very very recently, my only foray into this culture was by way of “Friends Reunited”. I have (and am proud to admit it) avoided the need to do “Myspace” or “Facebook” or (heaven forbid!) “Tweet”. I work all day with IT – mainly designing, writing and testing code in a variety of languages, and cannot think of any event during my working day that would warrant any form of continuous updating of my life to the general public. Who in their right mind would be interested in it anyway (other than my pay masters – and i suspect that if they saw a continuous stream of twitterings or other nonsense during the working day then they’d be inclined to worry about my delivery deadlines rather than how my day has gone).

However it would appear that the rest of the world moves on, and these things are ‘in’ and popular, and people now regard me as being eccentric for not joining in with the general “2.0″ fun…Even my family appear to be heading down this slippery slope. I know for a fact that both my Wife and Son are “face-booked” (my term, i’m sure there is a proper term but just humour me please:) ). Beyond my immediate family i am aware that others within the blood-line are busy filling their days with all this stuff.

Am I worried? No. In fact i’m reasonably happy to let this sort of thing slide by – i have other things to do with my life. However as a token gesture to it all i have succumed to blogging. The Coopers Consulting Solutions site is back in action (having spent the last 2 years as a static page) and i have a large collection of notes and other information that will eventually get back-ported onto here.

And i’m not just talking about blogging as a business tool either. I have also finally gotten round to setting up our home website with Word Press, and configured personal blogs for both Wife and Son, to do with as they please. I too may well join in the fun, although i’m more inclined to play the voice of the Rabbits, Cats or even the Chickens that we own, than actually convey any information about my own life.

So you see my problem?

On the one hand, i refuse to join the Web 2.0 generation fully and properly, and use the frequent line of “I don’t have the time to indulge”… and on the other hand here i am blogging away on a Saturday evening (whilst waiting for the WP set-up scripts for the family blogs to complete in the background).

Anyway, enough of this guff. Normal service will be resumed in due course, and the content of articles published here will revert back to business.

Thank you for your attention – no go away and get a life

Phew!

So, after a very long weekend’s work, i was able to open up the HRMS Aces forums and send out a bulk email to all registered users to say it was all up and running again.

Was it worth it (i hear you ask)?  Hmm let me ponder out loud on that one.

To the untrained eye, there is very little difference between what the site looked like on Thursday 9th April and what it looked like when the doors reopened (late) on Monday 13th April.

Sure the main site URL has changed from www.hrmsaces.com to www.hcmaces.com, but i’ve set-up redirects to send everyone over to the correct place should they forget. This is a distracting change – in order to keep the downtime to a minimum i had to have a domain in place to receive the site before i could tinker with the DNS entried of the old domains. In due course everything will be shoved back under HRMSACES.COM and the other sites will be set to point to it. But i digress…

The eagle-eyed may spot that the version of the forum software (ubbthreads by the way, in case you were asking) has changed from 7.4.2 to 7.5.2, but again this is subtle – the version change effectively implemented a large number of bug fixes, and some minor functionality changes which have not yet been activated…The day-to-day bread and butter of the site is unchanged.

So what has changed to justify the 30-40 hours worth of work that i put in over the Easter weekend? In short, nothing! Nothing as far as Joe Public are concerned. And in theory that should be the way its done.

The site has been moved to a new server, run by different hosting company, under a new domain, owned by a different registrar!

If you want a more familiar analogy then think of it this way: On Thursday you banked with Lloyds TSB (say); had done for all your life. On the following Tuesday morning you wake up and now bank with HSBC (say), and all of your direct debits, standing orders and other 3rd party agreements (credit cards etc) have all been seamlessly moved….That is effectively what was done.

It was not easy. It relied on the fact that i’d packed my family off on a camping holiday during this period so that i could work over the weekend without guilt, and also relied upon the support guys at my new hosts being around to answer some of my support calls during this time. It also relied on about a month’s worth of planning in advance to ensure things were in place to go through with it.

Would i do it again? No probably not. However it needed to be done, and i’m glad that its completed.

Now i can get on with other juicy internal IT projects that i’ve been putting off for too long

Today’s the day.

The day i’ve been planning (and dreading) for almost a month now, has finally arrived.

I get to pull the plug on HRMSACES and move it lock-stock across to a completely new hosting company. New server, new hosts and a new domain.

The idea of switching hosting companies came up in a conversation with an old fried at the UK Oracle User Group meeting in Birmingham last December.

I’d met up with an old DBA lag of mine, and whilst the free beer was flowing we discussed many pressing issues of the day (state of the contract market, temperature of the beer, Digital SLR’s, Web Hosting, Apple Mac’s etc). Web Hosting got me talking. At the time i was using 1and1 internet as my primary hosts (and registrar) for all of the domains i own, including the HRMSACES forum site. However i was (and still am at present) paying for a professional hosting package, which chimes in at £29 per month plus other sundries for the numerous additional domains held. This means i am paying, on average, over £70 per quarter.

Tim (the DBA) had almost choked on his beer at that (almost but not quite – a DBA defends his beer to the end and would rather miss an RMAN backup than see beer go to waste). Further discussions ensued, and the outcome was that i was being fleeced and should mosey over to an outfit called Fuzioned, who would kit me out for significantly less of my hard-earned ££££.

So that is what i did…Actually that wasn’t. What i did was procrastinate for about 3 months before finally approaching them.

Anyway – the rest is a ramble for another day. What we have happening today is the “big one”. Having already moved over my own company web sites and email (with zero downtime i hasten to add) today we embark on a journey to move the (reasonably) active HRMSACES foums.

I suspect it won’t be smooth, and will take more time than i have planned, but hopefully it should all be right in the end…