It was a no-brainer then to dump Microsoft altogether and move to Ubuntu permanently, my distro of choice. Except for a small issue with sound not working, the Ubuntu installation was virtually painless. The ThinkPad will continue to be my Java development box, and while I still do Java dev work on Mac, I prefer to use it strictly for Objective C and Cocoa.

That’s not to say I’ve left MS for good. I still run XP on my Mac via Parallels, because I enjoy C# and Quicken runs best on Windows. But it’s looking like it’s only a matter of time before I move completely away from MS. Am I the only one? Hardly, and I’d be surprised if Microsoft weren’t concerned. They should be.

However, I now have a green, 1-pixel wide, vertical line running up the right side of the LCD that appeared out of nowhere one day. That I can actually live with; the random reboots, I cannot.

Still, I have a 3-year warranty on Mac, and I’ll be taking it in the first of the year sometime to get this minor annoyance taken care of.

After ruling out database and application server issues it was noticed quickly that one of our GVPs was generating an awful lot of logs, in particular, I was seeing a ton of VXML errors in the popgateway logs from two separate JSPs. Example:

VxmlUtil.cpp:3481 C=12:L=1:U=321 VXML Error: event(error), message(), element leaf_Document->vxml->[OfferVCB] form1->[VCBChoice] field1->prompt2
@http://orappr5.example.com:7777/VCB/pages/OfferVCB.jsp), current url (http://orappr5.example.com:7777/VCB/pages/OfferVCB.jsp)

NetMgt.log was also reporting problems:

TrapInterface.cpp:651 C=5:L=2:U=0 Trap [cnAsrOsrRzCreateError] filtered out!
TrapInterface.cpp:578 C=5:L=2:U=0    TRAP: cnAsrOsrRzCreateError
TrapInterface.cpp:579 C=5:L=2:U=0    OID:  .1.3.6.1.4.1.3814.29.1002
TrapInterface.cpp:619 C=5:L=2:U=0    wdProcID:  3820
TrapInterface.cpp:619 C=5:L=2:U=0    wdProcName:  popgateway1
TrapInterface.cpp:619 C=5:L=2:U=0    cnTrapSeverity:  3
TrapInterface.cpp:619 C=5:L=2:U=0    cnResellerName:
TrapInterface.cpp:619 C=5:L=2:U=0    cnCustName:
TrapInterface.cpp:619 C=5:L=2:U=0    cnApplName:  GenesysCSA
TrapInterface.cpp:619 C=5:L=2:U=0    cnSessionId:  {CD153988-3C08-4CFF-98D9-677244B501E4}
TrapInterface.cpp:619 C=5:L=2:U=0    cnTimeStamp:  Mon Dec 01 12:20:45 2008
TrapInterface.cpp:619 C=5:L=2:U=0    cnTrapDesc:  Error creating Server Session object. SWIrecResourceAllocate() failed.
TrapInterface.cpp:624 C=5:L=2:U=0    MESSAGE-STRING: TRAP=cnAsrOsrRzCreateError&OID=.1.3.6.1.4.1.3814.29.1002&wdProcID=3820&wdProcName=popgateway1&cnTrapSeverity=3&cnResellerName=&cnCustName=&cnApplName=GenesysCSA&cnSessionId={CD153988-3C08-4CFF-98D9-677244B501E4}&cnTimeStamp=Mon Dec 01 12:20:45 2008 &cnTrapDesc=Error creating Server Session object. SWIrecResourceAllocate() failed.&

It turns out that power had been lost in the server room that housed the GVPs over the weekend and while all seemed well after power was restored, the problem reared its head when call volume spiked Monday morning. While GVPs 1, 3 and 4 were operating normally, something became hosed on GVP 2 when the system unexpectedly went down.

A warm reboot of the box containing GVP 2 resolved things.

Since we’re due an upgrade to 7.6 in just a few days, I’m not sure what more root cause analysis can be performed. It’d be nice to know what really caused this issue, but we may never know.

For those of us using the Mac client, and this seems particular to Mac OS X 10.5.5 users, there is now a very noticeable and quite annoying flicker during game play, especially when in towns and cities, within structures, or around trees. I have noticed this with previous patches, but it was the rare occurrence. However, with Echoes of Doom, it’s happening all the time. Drives one nuts, it does.

Fortunately, there is a fix.

Exit the Mac client and edit config.wtf, which is located in the WTF directory in the World of Warcraft installation directory. Add the following line to the end of the file:

SET fixedFunction “1″

Save, then restart the client. No more flickering.

Update 22 October 2008: Looks like the additional setting does not work. However, I’ve found another solution, and so far so good, I’ve played a few hours with zero flickering. Prior to the patch, I’d been running in full-screen mode. To stop the flickering I switched to Windowed Mode, which you can find in the Resolution settings under Video options.

Additionally, checking Maximized basically gives you the full-screen mode feeling.

I’m crossing my fingers hoping this finally gets rid of the annoying flickering and flashing once and for all.

I had just launched Visual C# when Mac froze for a moment, Parallels crashed, and the screen wiped gray with this lovely dialog box in the center:

Upon restart, Mac mocked me further:

As if I needed to be reminded.

Parallels is useful, but it sure blows sometimes.

We access dozens and dozens of stored procedures from our EJBs and have never had this issue before. The SP tested fine, it wasn’t that, so I decided to look at a few other SPs we execute and see what was so different about this one. What I found was while we normally handle SPs that return a cursor or an integer, for example, this was the first time we were trying to implement a SP that returned a boolean value. I didn’t think that was a big deal, but with further research, it turns out it is.

Per Oracle here, BOOLEAN return values are not supported:

It is not feasible for Oracle JDBC drivers to support calling arguments or return values of the PL/SQL types TABLE (now known as indexed-by tables), RESULT SET, RECORD, or BOOLEAN. There are currently no plans to change this. Instead people are encouraged to use RefCursor, Oracle Collections and Structured Object Types.

As a workaround, you can create wrapper procedures that handle the data as types supported by JDBC.

I did not know that.

But now I do and with a wrapper function in place, my EJB is happy again.

Caching causes us other problems, though, typically in our development environment, where it may not be a huge deal but is annoying nonetheless.

<rant>

Take for example deploying applications. This snippet from our application server log shows what can happen:

javax.naming.NamingException: Error instantiating web-app
JNDI-context: No location specified and no suitable
instance of the type 'com.example.SomeRandomEJB' found
for the ejb-ref ejb/SomeRandomEJB

The exception has absolutely nothing to do with the bean I was actually working on, it’s just some goofy caching weirdness occurring during redeployment. It is easily resolved by bouncing the container. Which apparently refreshes Oracle’s caches. This happens maybe 1 out of 6 time, it’s a nuisance and a headache. I’ve learned to deal with it.

</rant>

A production server move had occurred the night before, so it seemed obvious the problem had to be related to that. There was an IP change with the move, but we didn’t have issues pinging the box by name or IP, so we knew the change had propagated across the network.

A test app running on a development app server proved that the server move didn’t affect that connection, so we knew it was only the production app servers having this issue.

The consensus from everyone seemed to be that the application server was somehow caching the old IP address or was still using an existing connection that was no longer valid. I’m no expert on the OAS, so who knows. Turns out that all we needed to do was bounce the OC4J_Wireless container on both middle-tiers, and problem solved.

When in doubt, sometimes it’s just best to start by bouncing the container. Our DBA is investigating…

Updated to add: From an application deployment standpoint, we bounce the containers all the time, it’s nothing new for us. On our development servers, sometimes apps work immediately after redeploying without a restart, sometimes they don’t. On production, it’s our policy to perform a bounce after redeployment, no matter what. I seem to recall running across Oracle documentation years ago that recommended a container restart in redeployment/deployment situations. Such is the nature of caching, I guess, and now we have another situation where container restarts are proving necessary.

Overall, I thought it was a good effort, but certainly not the best. On the negative side, I found a surprisingly large number of grammatical and spelling errors. That always irks me when it comes to technical books, since for the price you are paying, you’d expect a much better effort at proofreading and editing. Also, while the book claims you don’t have to have prior programming knowledge, my guess is that certain OO concepts might be difficult to grasp at first unless you’ve encountered a language like Java or Smalltalk previously. However, if you have been doing any kind of object-oriented programming, you’ll get up to speed fast with Objective-C. And that’s another positive for this book, it’s a quick read. Kochan doesn’t waste time and gives you good examples right out of the gate. Don’t expect in-depth coverage, however; it is, after all, only an introductory text.

I’d definitely recommend the book if you have prior programming experience. If you don’t, I’m not sure this is the best book to start with (of course, I wouldn’t know what else to recommend in its place.) Also note that the current version DOES NOT cover the current version 2.0 of Objective-C. The good news, though, is Kochan’s update is set to release at the end of November, 2008. Wait until then.

Up next: Cocoa: Programming for Mac OS X.

However, Mac did perform a random reboot tonight, which when it happens is a bit startling. Thinking back, I believe it’s done it one other time since I’ve owned the machine. In other words, quite rare.  I can’t recall what triggered it last time, nor am I really sure what happened this time, but I’m leaning towards pointing the finger at Parallels, since I was poking around with Office 2007 and its PDF/XPS plug-in when the reboot occurred.

Parallels will blow every so often, courtesy of Windows I gather, so I shouldn’t be the least bit shocked. But I only ask Windows Parallels to contain its Windows blue screen of death to the virtual machine only, and not infect Mac.

Pretty please?