Raible Designs

How do you determine a good MaxPermSize?

I know I'll probably get beat up for not knowing my JVM Turning parameters. I admit that I should know them better than I do. Hopefully this post will help us all understand them a bit better.

Ever since I upgraded appfuse.org to AppFuse 1.9.1, it's been experiencing OOM issues. They've been so bad that the site is lucky if it stays up for more than an hour. I've done a fair amount of performance testing on a single AppFuse application (and gotten very good numbers), so I was pretty puzzled by the whole situation.

To reproduce the problem, I downloaded all 5 demos to my machine and began profiling with JProfiler. Nothing stood out, but I was able to reproduce the problem by clicking through all the different applications. While testing, I had my JAVA_OPTS set to -Xms256M -Xmx384M.

After staring at JProfiler for hours, I gave up and sent my findings to the AppFuse mailing list. After going back and forth with several ideas, Sanjiv came up with the winner.

Did you try increasing the max perm size (-XX:MaxPermSize=256m)? Max Perm size is running out of memory and not necessarily the main memory. Class metadata stuff is placed in the perm memory (google for more details) and since we're using Spring, Hibernate and Tapestry which all use a lot of reflection, proxying etc, it's not surprising that max perm size is running out of memory.

Based on his advice, I added -XX:MaxPermSize=256m to my JAVA_OPTS, fired up JProfiler/Tomcat and began hammering my local instance with WAPT. 15 minutes later, with 20 simultaneous users, the heap and memory were humming along nicely with no issues. I made the change on appfuse.org and it's been up every since.

This experience has motivated me to start adding "-XX:MaxPermSize=256m" to all my JAVA_OPTS. Is this a good idea? If so, is 256m a good value to use? If not, what's the best way to determine (or guess) the proper value for this setting?

The T2000 Arrives

I received my T2000 server from Sun yesterday, and now it sits in its original packaging in my basement. I may not even unwrap and install it. When I originally ordered the server, I heard you could blog about it and possibly win a free one. Since then I've heard different; there's a contest with many entering, and they're only giving away a couple of them. Because of this, I've somewhat lost my motivation to install the sucker. Partly because I know I'll have to send it back, and partly because it doesn't look that fast. Regardless, it's nice to have Jeff's setup instructions if I do decide to assemble/install it.

Right now, my schedule is packed full of client work - all the way until mid-June. So if I'm going to build this bad boy, it'll have to be as part of my "late night" schedule. Right now, that's booked with Spring Live, designs for the CSS Framework and AppFuse/Equinox development.

To further justify my lack of motivation, I don't think the T2000 will solve my OOM issues with CruiseControl and Ant. I've been talking with one of the Ant developers, and it appears to be caused by the <copy> task in Ant. With any luck, I'll find some time to setup CruiseControl and do some performance testing of web frameworks before the T2000 goes back in mid-June, but it doesn't look good right now.

Off to The Shop!

It's been quite some time since I last worked on the bus. Something about the spring weather around here motivated me to finish it up last weekend and get it ready for the shop. It only took a couple hours to pull off the remaining items I could, and then I called the tow truck. He showed up this morning, and now it's off to the shop for body work and paint.

With any luck, when it comes back it'll look like this one. I doubt it though - there's going to be quite a bit more work to do on suspension, electrical and interior. Estimated time at the shop: 6 months to a year. Below are some "before the shop" pictures I snapped.

Images powered by Lightbox JS v2.0 and Flickr.

Free Wireless along 16th Street Mall

First of all, I'm amazed that my Tips for Productivity and Happiness at Work generated so much traffic. It even ended up on digg.com for crying out loud. 800,000+ hits on Saturday alone. Yikes!

While today's story won't help you get disconnected and enhance your productivity, it might enhance your happiness at work. If you work in an office like I do, where you can work pretty much anywhere, you'll be happy to know that Denver's 16th Street Mall now has free wireless. That's pretty damn cool - especially since our offices are on the mall. Hat tip to Greg, who sent me this article in an e-mail.

Update: I forgot to mention, I'm pretty damn impressed that Roller was able to handle all the traffic this weekend.

Tips for Productivity and Happiness at Work

Every so often, I get asked what my strategy is for "getting things done". This morning, I had a short session of mass productivity, and on my ride into work, got inspired to jot down a few tips productivity tips. Keep in mind that I grew up in the back woods of Montana with no electricity and I'm mostly Irish.

• Write stuff down. The first and most important means to getting things done is to keep a todo list. I keep mine on Ta-da List and update it daily, if not hourly. My two most important lists are "When I have time" and "This Week". I regularly re-order "This Week" for the current day.
• Quit reading e-mail and blogs. One of the ways I can tell I'm in uber-productive mode is my unread (or starred) mail piles up and I haven't read any blog posts (or blogged myself) in a couple days. I like GMail because I can easily star an e-mail after briefly skimming it. For some reason, flagging e-mail in Thunderbird or Mail.app doesn't work for me - I rarely go back and find those e-mails. With GMail, I'm always conscious I need to revisit starred messages.
• Work on open source late at night, with a beer on your desk. While I do get the opportunity to work on open source at my day job, I still find that I'm most productive at night. Maybe this is because no one bugs me via e-mail or IM, or maybe it's just because the world is asleep. The strange thing is I often find myself motivated at 3 p.m. for my 11 p.m. workload. However, when I get to 11 p.m., I'm not motivated to work on anything. I've found that cracking open a beer at 11 when I start helps me focus and quit worrying about all the other computer-related tasks I need to do. Also, on beer #2 or 3, you'll start to forget what time it is and really start getting things done. NOTE: this isn't for rookies. If you're a lightweight and get hammered on two beers - just go to bed when you start coding with one eye shut. It helps to sleep in the next day after doing this. If you finish off a six-pack before going to bed, it's probably best not to check your code in - you're probably going to spend the next day fixing it anyway. Regardless, this is a great way to get started on a new feature because you're less concerned about the details and more concerned about the big picture. I've also found that "bug fixing juice" can be great for fixing bugs - it gives you a different perspective on the problem. I told you I was Irish didn't I? ;-)
• Work disconnected. To further facilitate not checking e-mail or reading blogs, I've found that going to a coffee shop w/o connectivity is my most productive environment. They have liquid motivation in the form of coffee, and you can feed your brain with breakfast/lunch or some kind of snack. My most productive days are the ones where I show up at my local Einstein's (bagel shop) at 6 a.m., have two cups of coffee, and work with my headphones on. After the coffee and uber-productivity, I often have an awesome ride to work and barely notice the miles. NOTE: I've found that I'm more productive writing code late at night and authoring articles/books in the early morning.
• Listen to music while you work. Some noise-cancelling headphones and your favorite music can do wonders for your productivity. Of course, earbuds work just as well - whatever makes the music sound good. Good music can really help you "get into the groove" of what you're working on, regardless of whether it's writing or coding.
• Work long hours on Monday and Tuesday. This especially applies if you're a contractor. If you can only bill 40 hours per week, working 12-14 hours on Monday can get you an early-departure on Friday. Furthermore, by staying late early in the week, you'll get your productivity ball-rolling early. I've often heard the most productive work-day in a week is Wednesday.
• Avoid meetings at all costs. Find a way to walk out of meetings that are unproductive, don't concern you, or spiral into two co-workers bitching at each other. While meetings in general are a waste of time, some are worse than others. Establish your policy of walking out early on and folks will respect you have stuff to do. Of course, if you aren't a noticeably productive individual, walking out of a meeting can be perceived as simply "not a team player", which isn't a good idea.
• Sleep. While working late nights can be productive in the short term, doing it consecutively will burn you out quickly. Getting a good night's sleep can often lead to greater productivity because you're refreshed and ready to go.
• Work on something you're passionate about. If you don't like what you're doing for a living, quit. Find a new job as soon as possible. It's not about the money, it's all about happiness. Of course, the best balance is both. It's unlikely you'll ever realize this until you have a job that sucks, but pays well.

To follow up on that last point, I think one of the most important catalysts for productivity is to be happy at your job. If you're not happy at work, it's unlikely you're going to be inspired to be a more efficient person. Furthermore, if you like what you do, it's not really "work" is it?

Here is some general advice I give to folks about jobs and careers. I realize that I'm biased here because I have a good job and I've been fairly successful in my career. However, I also grew up with virtually nothing, and learned all my computer skills on my own. Therefore, I believe that anyone can be successful (meaning: happy with that they do, and financially stable at the same time) if they put their mind to it.

• If you're not happy at work, quit. I got into contracting early-on in my career (6 months out of college), so I got spoiled with good rates early on. It was also the late 90s, so you could easily switch jobs if you didn't like the one you were in. However, I've always had the policy that if you hate your job for more than 2 weeks, quit. It's just not worth it. Some people enjoy bitching about their jobs and complaining about their co-workers, so this doesn't apply to everyone. However, if you're truly miserable - quit. I've done this a couple times, and it's always led to better opportunities (even if I had to wait 3 months). Julie's done it once or twice and it's always worked out for the better. Once she even got a 20% raise by quitting her traveling-consultant gig with KPMG and going back to her old company, Qwest.
• Always try to be a contractor before a full-time employee. This especially applies to younger folks who don't need the security of insurance for the family, stock plans, etc. Contractors typically make 50% more than full-time employees and tend to be excluded from company politics that might make the workplace difficult to deal with. While you won't get billable vacation time, you will get the freedom to take however much vacation you want - as long as you get your stuff done. The higher rates can generally make up for the unbillable time while on vacation. However, if you travel too much for fun (or conferences, etc.), it's likely your year-end salary will equal that of a full-time employee. If you make good money as a contractor, you'll also get the opportunity to start your own company (to save money on taxes). This can be a great learning experience. The biggest fear that folks have about "going independent" is they'll have a hard time finding their next gig. If you're productive and blog about what you're doing, this shouldn't be a problem. I haven't had an "interview" since 2002 and haven't updated my resume since then either. Networking at your local JUG and conferences is key.
• Don't work at a company with a two-week vacation policy. I've never worked at a company with a two-week vacation policy, and I hope I never will. The two full-time gigs I've had in my career have had no vacation policy. This is usually only found in startups - but it generally amounts to "get your shit done, and you can take off all the time you want". My parents both work for the BLM (government agency) and they get 6 weeks off a year. Furthermore, they can earn "comp time" (more days off) by simply working overtime. If you have the ability to take time off whenever you want, it'll likely lead to you being more motivated to work long and hard - b/c you know when you finish that project, you're heading to Cancun for a week.
• Don't travel if you have kids. If you're single, traveling for work is pretty cool. New places, new people to meet - and seeing the world can be very cool. If you have a spouse, it's likely your desire to travel will decrease, but it's still not that bad. If it helps your career, it's probably a good move. It also helps to save money since everything you do is generally expensible. When Julie and I first met, she traveled 100% and made $20K more per year b/c of it. When you have kids, everything changes and traveling sucks. You miss their first steps, and when they get to toddler-age, they'll want to go with you. Hearing your daughter say "Daddy, can I go with you?" can be heart wrenching when you're leaving on a Sunday afternoon to spend a week with people you've never met before.
• Ask for more responsibility. If your job sucks, but the company is pretty cool - you may want to ask for a shift in responsibilities. I was once an HTML Developer at a .com. It sucked because we were constantly waiting on the Java Developers to fix bugs we found in their code. Finally, I got tired of waiting and asked my boss to show me how to fix the Java bugs. Granted, I wrote some pretty horrendous code at first, but my boss and co-workers helped a lot and w/in a year I was doing 80% of the Java Development.
• Most things can be learned by reading. If you want to learn something new (for your current or next career), the best thing to do is read. The world's knowledge resides in books and you can learn a lot. Of course, the best way to retain that knowledge is by doing, but reading is a great first step.

Those are my tips, many of them off the top of my head. I may add more as time goes on - but hopefully this helps in the meantime. Please share yours if you have any.

First Avalanche Game

Last night, my mom and I took Abbie and Jack to their first Colorado Avalanche hockey game. Jack's a huge hockey fan - has his own stick and everything - so we figured they'd both enjoy it. We had nosebleed seats (2nd row from the top), but the kids definitely enjoyed the game. 6 goals in the first period probably didn't hurt. The only problem was they kept screaming "Go DU!" - much to the chagrin of the all-too-serious Avs fans around us.

Parallels Workstation for OS X fails the 10 minute test

I gave the Parallels Workstation a run for its money on my MacBook Pro this evening. It failed the 10-minute test because it took 45 minutes just to install Windows. Beyond that, the networking doesn't work and you can't full-screen the application window. It's not included in the free beta download. I think I'll just continue using a 2nd Windows box instead of making my MacBook Pro a 2-n-1 machine.

Update: The Beta 2 and Beta 3 solve these issues. However, I'm unable to install WebLogic 9.1 on either of these. I get a nice "could not extract the archive since it is corrupted. Error code = 2" error from the installer. Luckily, I found this howto for OS X. The same Windows installer works fine on a real Windows XP box.

Getting ready for the Sun Fire T2000 Server

Like Jeff, Bill and Bruce, I signed up for the free 60 day trial of the Sun Fire T2000 Server. Why? Because I heard you can get one for free if you blog about it enough. My plans are to setup continous integration environments for AppFuse and Roller using CruiseControl. I also hope to do some performance tests b/w Java persistence and web frameworks. Finally, I'd like to some single-server vs. clustered server performance tests using Tomcat.

When we built our house way back in 2004, we had it wired with fiber. Even though we still aren't using the fiber, I also had them install ethernet throughout. Today, I finally took advantage of it. I moved my cable modem downstairs and hooked up RJ45 connectors on both ends, so I can now plug my office network into an ethernet outlet on the wall. Good thing my Dad is a network guy - he made it all pretty damn easy.

It was a great afternoon in Denver: 85 degrees, cold Fat Tires in the fridge, kids running around playing in the kiddie pool - and I got my house re-wired in under 30 minutes. The "old basement" will soon become a server room. Now I need a rack - for the T2000 as well as an old Dell Dimension Fedora box. Any suggestions?

[ANN] AppFuse 1.9.1 Released

This release includes improvements and upgrades to Tapestry 4.0.1, WebWork 2.2.2, as well as support for using AppGen to reverse engineer database tables (using Middlegen). iBATIS is now supported by AppGen and a Create DAO tutorial has been put together for iBATIS. iBATIS and Middlegen support were provided by Bobby Diaz - thanks Bobby! Also, a big thanks goes to Mika Göckel for writing an XFire Tutorial and installer. To install and configure AppFuse for development, see the QuickStart Guide. Thanks to all the sponsors who have contributed products and free hosting to the AppFuse project.

• Download
• Release Notes
• Learn more about AppFuse

To see how AppFuse works, please see the following demos (username: mraible, password: tomcat):

• Demo of Features in Flash (needs to be updated, things are much easier now)
• Struts + Spring + Hibernate
• Spring + iBATIS
• WebWork + Spring + Hibernate
• JSF + Spring + Hibernate
• Tapestry + Spring + Hibernate

Comments and issues can be sent to the mailing list or posted to JIRA.

What happened to Middlegen?

In AppGen (AppFuse's code generator), I recently integrated support for generating CRUD code (and tests!) from database tables thanks to APF-95 from Bobby Diaz. Previously, it was only possible to generate code from POJOs, but now both are supported. The "mapping files from tables" is done using Middlegen.

Unfortunately, Middlegen seems to be a dead project, especially since its main web site is missing (fixed!). I'd love to use another project that generates XDoclet tags in hbm.xml files, but I don't believe there is one. AFAIK, the Hibernate Tools project does not support XDoclet with its tools. Anyone know where the Middlegen documentation is located? All the documentation used to be on their site.

The good news is the solution I have works, and it works well. However, I definitely look forward to re-writing AppGen. XDoclet as a templating language sucks, and reading XDoclet tags to determine how to generate the UI seems pretty hackish. Hopefully we can use APT and FreeMarker (or Velocity) to process annotations and spit out code in a future release.

Update: The Middlegen website was lost, due some heavy server upgrades. The problem has been fixed and it's back up and running. Now I don't feel so bad about using Middlegen since the project appears to be alive and well.