Rainy Bugfix Wednesday June 1st, 2005

No power this morning until a few minutes ago. Light rain but no thunder. Surprised that the power went out.

Feeling a little better this morning, but as the days generally go lately I start out focused and then quickly defocus throughout the day. By mid-afternoon I find myself having made little progress on the 50 fronts I've taken on and then my motivation wanes. In the past I fought this with little prioritization checklists and I just worked down them. I'd prioritize in the morning and work from there throughout the day. I'm doing that these days but I still get defocused. I think it has to do with the amount of potential that I'm seeing out there now. For years I was kind of working off in a corner, believing in what I was doing but never seeing anything out there related to it. Now there's this explosion of activity on microcontent but I don't have the time or resources to get to all of it. Time for funding? Prolly.

Update 10:01AM: A little motivation in the RSS feeds this morning:

From Mark Cuban, owner of the Dallas Mavericks and co-founder of broadcast.com: In basketball you have to shoot 50pct. If you make an extra 10 shots per hundred, you are an All-Star. In baseball you have to get a hit 30 pct of the time. If you get an extra 10 hits per hundred at bats, you are on the cover of every magazine, lead off every SportsCenter and make the Hall of Fame.

In Business, the odds are a little different. You don’t have to break the Mendoza line (hitting .200). In fact, it doesnt many how many times you strike out. In business, to be a success, you only have to be right once.

One single solitary time and you are set for life. That’s the beauty of the business world.

Which is nice to hear. I'm right in my concept (datablogging, microcontent, structured blogging) but failing at my implementation right now. I think you have to be right once and you have to implement once... on the same project. Being right with a concept, failing to implement it, choosing another crappy concept and then implementing it well won't get you anywhere. You have to have the concept and the implementation together at least once.

Update 11:01AM: On the trail of the elusive database lock issue that's been causing me much stress for a while now. Fingers crossed I can solve this Xxxking bug and get on with life. It's been draining me for a while now. Last night, during a moment of self-reflection, I realized just how much this bug has been, well, bugging me. It's disconcerting knowing that the site that represents six years of your effort may go down any moment. And it's even more disconcerting realizing that you, despite writing every damn line of code, have no idea what's causing it. This has been a background stress on me for a while. But hopefully I'm on the trail this morning.

Update 11:05AM: There are 36 occurrences of "SELECT DISTINCT" in the code... surprisingly few. And 87 occurrences of "DISTINCT". Since I build a bunch of SQL dynamically I'll work down the list of 87 and see if I can find the offending code. I have to believe it's in some of the dynamically-built code where the variables to the left and right of the dot fail to bring anything to the table. We'll see.

Update 11:12AM: Found the offending dot, but not sure if it'll cause the issue. Good and bad news. Need more testing. Seems I'm hot on the trail though. I'm not going to fix the syntax yet... I need to see what that code does and then duplicate the issue on local testing machines. Once I do that then I can make some changes to the code and verify the fix.

Update 11:31AM: A little good news. The code where I found the issue is only used by the log type list functionality. And that functionality was launched right about the time that the issue started appearing. Again, I seem to be on the right path.

Update 11:35AM: Ok, a little bump in the road. Even though I know where the problem is, I can't duplicate the issue. I'll stay on it.

Update 12:01PM: Looks like it isn't a bug as much as it's a performance issue. The code doesn't cause a hard lockup as much as it just gets incredibly slow with the amount of data in production. I tested on one of the instances with as much data as there is in production and, sure enough, it locked it up. Which is a good thing. I can now duplicate the problem and I just have to fix it.

Update 2:06PM: Hot damn. It's fixed. Instead of going to the database I used the in-memory objects. A million times faster. Sweet! I'll fix a few other things, do some more testing and then launch to production.

Update 4:55PM: Finished the code. Uploading build right now.

Update 5:05PM: Still uploading. I put up the system message warning all reger.com users that the system will be shut down for an upgrade in 15 minutes so that they can save their work.

Update 5:25PM: Well, the build is deployed. I won't actually know if this was the only thing in the system bringing it to its knees, but at least it's one of them. Fingers crossed that all is well for a while.

Update 11:32PM: Went to the MIT Enterprise Angel Investors thing. Eh.

Been blogging a little about training, doing some reading. Nice to get away from the code a bit. I swear the only reason I'm doing this is because I (think I) solved the MySql lockup bug. It frees my mind a little. Maybe I can get to the 122 emails that I've eartagged for a response.

Heather's watching morbid crime investigation stuff on tv. I hate that crap. Too much focus on the absolute worst that humanity has to offer.

Haven't gotten any input from the Tomcat open source mailing list. Posted a question a couple ways and nothing has come back. I've said it before, I don't know why people don't generally respond to my questions. I've tried to ask the same question in a short, succinct manner and also in a completely detailed manner. Odd.

I'm not too tired, but I wish I was. I want to get to the gym early tomorrow. We have a baby appointment tomorrow evening at Northside hospital so I'll be working out at some point during the day. But I had a great sleep last night and the power was out so the alarm didn't go off. I should probably head upstairs and do some reading.

Update 11:41PM: Haven't eaten much the last two days. Something of a detox after some horrible eating the last few weeks. I've had 1800 calories total in the last two days and I don't feel even remotely hungry. Food still in my intestines is fueling me. Contrast that to a few weeks ago where missing 200 calories in a four hour period would send me searching for food. I've gained something like seven pounds in the last few weeks. I won't know the actual number until the food in my stomach/intestines clears out and my morning post-pee weight stabilizes. That generally takes three to four days. We'll see.

Update 1:02AM: As you can tell, I didn't go to bed. I got a few emails out. Not exactly tired right now. Surprising because the last few weeks I've been so frustrated with that lockup bug that I exhausted myself by the early evening. I feel like I've been freed. Jeez. Now watch, tomorrow it'll lock up again.

Update 2:20AM: Getting a little tired now. Heading to bed.