Oracle Open World: Day One Summary
http://www.oracleappsblog.com/index.php/weblog/oracle-open-world-day-one-summary/

Oracle Open World update #6: WebCenter Released!
http://www.it-eye.nl/weblog/2006/10/24/oracle-open-world-update-6-webcenter-released/

OOW 2006: Keynote speech at Oracle Develop by Thomas Kurian by Lucas Jellema
http://technology.amis.nl/blog/?p=1383

More Oracle Database 11g Beta Details
http://blogs.oracle.com/otn/2006/10/24#a294

Worst practices when using Oracle JDBC Drivers - great presentation by Douglas Surber by Lucas Jellema
http://technology.amis.nl/blog/?p=1385

Announcing: Enterprise-Class Support for Linux
http://www.oracle.com/technologies/linux/index.html

Oracle Open World Day 4 - 11g Performance & Scalability Features
http://www.rittman.net/2006/10/26/oracle-open-world-day-4-11g-performance-scalability-features/

Oracle Passwords
http://thinkoracle.blogspot.com/2006/10/oracle-passwords.html

YouProbablyDontNeedRAC
http://www.miracleas.dk/WritingsFromMogens/YouProbablyDontNeedRACUSVersion.pdf

Recovery Design
http://oracledoug.com/serendipity/index.php?/archives/1108-Recovery-Design.html
http://oracledoug.com/serendipity/index.php?/archives/1114-Recovery-Design-part-2-Asking-the-right-questions.html
http://oracledoug.com/serendipity/index.php?/archives/1117-Recovery-Design-part-3-What-is-it-with-RAC.html
http://oracledoug.com/serendipity/index.php?/archives/1120-Recovery-Design-part-4-You-need-more-than-one-database.html

Anyone want an introduction to Java Server Faces? - the “Dummies Guide”
http://groundside.com/blog/GrantRonald.php?title=anyone_want_an_introduction_to_java_serv&more=1&c=1&tb=1&pb=1

Guide to free Oracle Utilities
http://www.dba-oracle.com/oracle_news/2005_3_16_guide_free_utilities.htm

Articles on Java Security at the OWASP Java Project
http://www.theserverside.com/news/thread.tss?thread_id=42748

Lots of things been happening recently, company has been growly nicely and we are getting pretty busy. We’re now doing some work on .NET consultancy but the primary focus will remain within the wonderful world of Java or is that JSE or are we now looking into J2EE or it is JEE. Whatever!

I’ve also signed up to do the M.Sc. in the Management of Information Systems in TCD. It’s just about 10 years since I finished up there and the brain was getting a little water logged so I reckoned it was time to go back to school. So new pens, paper and even a new schoolbag bought and I am ready to rock. I feel my mind expanding already and it is only the first week - the highlight for me was seeing the following cartoon which I know my friends will get a good laugh out of. Of course it is untrue but they’ll get the joke (I think!).

I have a few interesting techie things to get uploaded over the next few weeks - so hopefully you can check back soon. Until then.

As usually happens in these scenarios it was immediately classified as an problem infrastructure problem so work was done looking at network usage, http response sizes, performance of the client and many other red herrings. One pointer that showed up during the analysis though was that there was a significantly high page-in rate on the database server. Typically a hgh page-in rate is the result of a large number of new processes being invoked (unless accompanied by a high page out rate but that is for another day).

Once we uncovered that little detail the problem became alot easier to resolve - the obvious question was - what process is being created so frequently on the database server, a few candidates were ruled out and once a logon trigger was put in place we could see that a large number of short sessions were being created to the database from the application server.

After being reassured that all of the application code was connection pool aware and only used pooled connections I became interested in the connection lookup code - after poking around I could see that the connection was being fetched from the “location” jndi reference instead of either ejb-location or pooled-location. In OC4J 9.02 and upwards the preferred location to fetch connections from is the ejb-location so the code should have beean looking there. Unfortunately a little typo had crept in from one of the developers in a recent release and the configuration file specifying the datasources had a problem for one of the multiple datasources used by the application. A simple change to the config file and a restart and we were back in business.

The lessons from that issue were:

• Not all problems are infrastructure related
• A logon trigger is so useful in tracing application server behaviour it should be on all the time and logging to a table that is used to publish connection statistics
• Always get your jdbc connections in OC4J from the ejb-location

It is amazing to think back to the glory days and remember some of the amazing ways money was splashed around - my favourite still though is the company that was offering a BMW to all staff members who recommended a certain number of people who got subsequently hired. I think it was ebeon and they went bang before any BWMs got given out but still - it certainly improved the flow of CVs into them when good people were thin on the ground.

Of course the title of this post is mis-leading - if anything the market seems to be picking up again over here but we do not see the same amount of money being thrown around by companies on their staff. Recently Ireland has attracted a number of big names from the .com community and they have setup operations here, Amazon, Google and eBay are probably the three most well known and the three that probably are the biggest players in online commerce at the moment.

Haven’t spoken to anybody working there yet but I am wondering if they are doing any research and development over here - fueling the ‘knowledge economy’ that everybody keeps banging on about or it is simply the operations area that has moved over here (for grant/tax purposes?).

• Adding a read receipt - this is very important and I am going to beat you over the head with the contents of this mail once I know you have read it. The amusing thing is that people don’t seem to realise that you usually get prompted to confirm that you would like to send a read receipt.
• Adding extra people to distribution list - people who are unable to convey their message in text resort to the schoolyard tactic of ‘my brother is bigger than your brother’ by adding incremental levels of management to an email as the discussion progresses. I assume the implication here is that you will stop the conversation because you are scared of what one of the managers will do when they see your mail exchange.
• Hidden receivers - the telltale sign of this one is somebody who is not the receiver or on the CC list chimes in with a response and the mail has not been forwarded. Personally I don’t understand the whole BCC thing - if you are writing something and a person needs to see it what is the problem in everybody knowing who is involved in the discussion.
• Flamboyant signoffs - otherwise known as ‘killing with kindness’. In business related emails I usually sign off with Regards - and even that is only something I started doing recently. However I have seen some business emails signed with "My kind regards" or "Warm regards" - if you like the person that much why don’t you ring them or call to them face-to-face to arrange a date!
• No subject line - should be a criminal offence, just shows that the person sending the mail either does not understandwhat they are saying, or that they are too lazy to bother or that they want to make sure you delete the mail.
• The curse of the ten thousand disclaimers - as emails get forwarded from one person to another you get the usual disclaimers added to the end. Sometimes it gets to the stage where the amount of text in the disclaimers is bigger than the actual text of the emails. Wonder how much Internet bandwidth is wasted on these useless blocks of text (why not just include a link!)

If you know of any other good ones drop me a line - make sure to put a subject line in and sign it: "My warmest and kindest regards"!

Note that for RHES4 you need to make a small modification to the redhat-release file to make the installer think you are running RHES3 which is all that was supported orginally. The workaround is documented in the OTN install guide referenced above.

The final issue that I faced was getting the automatic start/stop stuff running properly under Linux - the Oracle Administration docs suggest you create a file called dbora which rsh’s into your server and executes the necessary commands to start to database and the listener. The problem I found here is that the standard installs of RHES do not enable rsh/rlogin (for good security reasons). You could work through the steps of enabling them but I found another doc on MetaLink which covers another version of dbora which su’s into the oracle account to run the necessary commands. This works nicely in my environment so I went with that.

I had to tailor the new dbora script to point to the correct ORACLE_HOME and I then edited the dbstart/dbshut scripts to have the correct path to oratab (/etc/oratab) - then everything worked nicely.

Now I’m just installing instantclient to se if I can connect to the VMware instance running the database….

Perhaps I will be permanently tempted away from my beloved JDeveloper which I have been forcing myself to like more and more over the past few releases. It is getting better but when you look at IntelliJ there is no competition for code editing, however if you are an Oracle shop it is hard to argue against JDeveloper, the 10.1.3 release that is coming looks pretty good also.

]]> http://pjboyle.revelate.ie/?feed=rss2&p=16 http://pjboyle.revelate.ie/?p=17 http://pjboyle.revelate.ie/?p=17#comments Mon, 04 Apr 2005 20:30:00 +0000 Administrator http://pjboyle.revelate.ie/archives/16-guid.html If you are looking for a cheap and cheerful platform to develop and test Oracle on Linux stuff you should look no further that a new distribution called CentOS. It is based around the RedHat Enterprise distribution and just has the RH specific branding removed to keep the lawyers happy. it is a similar idea to WhiteBox Enterprise Linux which is also based on the RedHat SRPMs but I get the impression that CentOS has a more active community behind it so they will be able to keep up with the RH releases as they arrive.

If you are installing Oracle 9i or 10g on Linux be sure to check out this excellent page containing the details on the things to watch out for - http://www.puschitz.com/InstallingOracle10g.shtml

http://madbean.com/blog/2004/30/ep3.html

Reminds me that I must dig out that link with the guy trying to be all serious pretending to be in StarWars - you know the green screen one. If you know the link send it on.

]]> http://pjboyle.revelate.ie/?feed=rss2&p=15 http://pjboyle.revelate.ie/?p=14 http://pjboyle.revelate.ie/?p=14#comments Thu, 17 Mar 2005 08:46:00 +0000 Administrator http://pjboyle.revelate.ie/archives/13-guid.html I had an interesting exchange recently with a developer who told me that unit tests are not developed on their project unless they are chargeable to their client. I was amazed but not surprised at the situation - I would have thought that during the development of quality work it would be assumed that you would be developing unit tests alongside the code.

I can see the point that these companies don’t give anything away for free (apart from advice that consists of them telling you that more of their consultants are required!) - but the development of Unit Tests should be assumed as part of any development project these days. For unit testing time not to be factored into the project estimates shows that many of these companies are not keeping pace with the field.

There is some debate in the industry about the granularity of unit tests and I agree - too many times I have seen silly tests that do the is 1+1=2 type stuff. Some developers look on the tests as a necesary evil to getting their module released. They just don’t see the benefit. I think that Unit Tests really pay for themselves during the refactoring work in a project - just having the confidence to get stuck in and fix some dodgy code, knowing that the test suite will keep you in line is a great boon. Many people shy away from refactoring code they don’t own since they have no idea of the ripple affects when it changes.

Anyway - back on topic - the next thing I will hear is that these big companies are charging a certain amount per email read and responded to……..