JavaBlogs ongoing fiasco
There’s something faintly comical about the total inability of anyone at Atlassian to write a java web app that manages to stay up more than a day or two.
JavaBlogs, has, since its very inception, been a delightful example of the ‘toss it out there, hope for the best, get bored, moved on’ approach that we’ve come to love and admire so. The early days were fraught with proxy errors, hours of downtime, and general unreachability with little to no warning, notice, apology, or explanation.
However, things got better, it all stabilised, I ran out of ammo. Help was forthcoming though in the shape of one Charles Miller, who let go of his last vestiges of sanity when he joined Atlassian. He took it in his head that it’s time to rewrite javablogs. Out with the old, in with the new! Let’s use a new persistence engine! Let’s use new API’s that get the blood pumping and the cvs updates flowing! Let’s bat at a variety of shiny objects dangling in front of us!
So we have javablogs reborn, mostly similar dysfunctional UI, some pointless graphs, and a return to our beloved instability.
The accusing finger of blame, more often than not, waggles in the direction of poor old postgres. Fine, you add your indexes, do a tuning dance, and all is well. Sadly, that seems to have had little effect. Almost daily, javablogs goes down now. The working hours of Atlassian can easily be inferred based on how long javablogs stays down. The average uptime, according to netcraft, is a pathetic 5 days. Particularly hilarious is that now and then, instead of the server simply falling over (proxy error), we get an OGNL error. I’ll avoid making snide remarks over what the actually means, however tempting it may be.
Of course, it’s not just uptime. The UI itself leaves plenty to be desired. To see your list of despised and ignored blogs, you have to click on ‘favourites’. The logout page has a link to ‘login’. Upon clicking on that, you’re given a one liner that says ‘scroll down and log in somewhere down there’. The graphs are meaningless, notifications of downtime non-existent, explanations non-forthcoming, and the content is, as ever, disgraceful.
July 22nd, 2004 at 12:55 pm
Makes JRoller look awesome then? ;-)
July 22nd, 2004 at 1:03 pm
So one assumes JRoller is now a shining example of uptime and good design?
July 22nd, 2004 at 1:21 pm
One of us is in deep draable. You ah mein naow, you belong tew mee, and I hope yew left enough rewm for my fist, because I’m going to RAM IT INTO YOUR STOMACH!
July 22nd, 2004 at 1:46 pm
Yup, somehow this time nobody else is raising a ruckus even though to my recollection this is the worst downtime experience ever with JavaBlogs. Crazy what we can come to accept and even defend.
July 22nd, 2004 at 2:27 pm
Oh, the irony is that fewer folks will read this bile exactly because javablogs is b0rked.
July 22nd, 2004 at 2:45 pm
Let me just say this:
July 22nd, 2004 at 3:00 pm
Is javablogs open source? If so, where is the source? I want to see how terrible it is.
July 22nd, 2004 at 3:52 pm
It’s not just JavaBlogs. I frequent many community sites running PHP or mod_perl, running ASP or ASP.NET, in addition to those running Java. And quite honestly, there’s obvious fragility in Java-driven sites. From JRoller (now it’s doing much better), to TSS, to JavaBlogs, to java.sun.com (down several days at the beginning of this year), to individuals’ sites that often greet you with a stack trace, like Hacknot and Ted Neward’s
July 22nd, 2004 at 4:38 pm
I’ve said it before, but there’s a lot more to running a web site than the code behind it. Clearly Atlassian doesn’t have the spare resources to make javablogs a 24×7 99.95% uptime community site, and they never promised us any such thing. It takes a lot to keep a heavily trafficked site up and running, no matter whether it’s Java, PHP, etc.
July 22nd, 2004 at 4:52 pm
Come on Jason, do you really believe what you say there yourself? It’s not like javablogs is a complicated ecommerce site? Over the many years that I’ve deployed web applications, they almost never required any maintenance and stay up as long as the reboots in between server security patches. In terms of true web application maintenance, a daily postgres vacuum is all that I ever needed. Maybe their application *is* far too complicated and doing way too much for its purpose. The site looks to me as though it’s run on a laptop and it’s going down when the owner takes it home with him … it’s a really sad example of a java web application.
July 22nd, 2004 at 6:04 pm
When I first clicked through to this article from Javablogs, I got a JRoller error. :)
July 22nd, 2004 at 11:08 pm
Ha ha yankees!!! for me and my antipodean waking hours, JAva Blogs is much more efficient, ‘cos Charles and co will come in during my waking hours and fix the error while YOU’RE all asleep.
July 23rd, 2004 at 12:56 am
OGNL error…. are they using Tapestry?
Wow, that’s kewl. Tapestry wrox!
July 23rd, 2004 at 8:51 am
Hani, i think you are sour about Webwork2. Reasoning, you keep workig for WW1.4. And you hate OGNL. And yes, i agree that they should never use OGNL in WW2, and that was a bigmistake to remove command interface in WW2 (it came back with 2.1 tough.)
July 23rd, 2004 at 9:20 am
Since the rewrite I think the webapp itself has been pretty well-behaved. Much of the last few months’ downtime has been due to kernel freezes on the rackspace server. The box is pingable, but other ports (notably ssh and http) hang after connection. After much fruitless arguing with rackspace, we’ve been looking for an alternative host. I think something may come of that soon.
July 23rd, 2004 at 11:29 am
From my brief conversations with the Atlassian guys, Jeff is correct. A majority of the recent downtime has been problem with the server – operating system and/or hardware. In order to better the situation, Atlassian is moving the server to a new hosting provider, namely us. We will be working with them to ensure that javablogs is available 24×7. Javablogs will be on a dedicated server with monitoring.
In my opinion, Javablogs is a great community service in which Atlassian has invested time and money. They continue to do such, not because they are required to, but because they choose to. It definitely has not been forgotten as this bile suggests.
July 23rd, 2004 at 2:33 pm
What’s really annoying about JavaBlogs is its insistence that all links come through its redirector. If you’re lucky enough to have grabbed a valid RSS feed from it (or via some other aggregator, like Yahoo!’s), you’d think that you’d be shielded from the JB downtime. But no! Every article points through http://www.javablogs.com/Jump.actionid.
Maybe if they didn’t insist on being in the middle of every single “click-through”, they wouldn’t have load issues. Yeah, I know that’s how they count the popular articles. I don’t know how they count the ones I can’t get to because they’re down. Heisenberg would be proud.Perhaps an RSS feed for those of us who don’t feel like being Nielsen families would be a nice compromise.
July 23rd, 2004 at 6:33 pm
hani if you have any self-respect left after hanging out with matt raible at javaone, you’d better get on it and bile that dumbass. you can’t bile people like bob mcwhirter and give this guy a free pass.
look at his stupid blog, he has the gall to have a sourceforge project that exists entirely of other people’s eclipse plugins he has zipped up and “released”. yes, the idiot even has release notes for it.
sympathy for morons who don’t know they’re morons is just not a good thing…you’ll feel better after you bile him.
July 24th, 2004 at 12:07 am
You need to check out gameblogs.org to see another approach to blog aggregation: simple php, blogs are categorised, you can view entries by your favourites, and there’s no registration, simply add a blog and you’re in!
July 24th, 2004 at 4:32 am
I agree that releasing a zip of Eclipse plugins is moronic. But what about the 2100 downloads from the prior release? Bile me – I probably deserve it for something or other. Car bombs maybe?
July 24th, 2004 at 11:18 am
I dunno, an aggregation of eclipse plugins isn’t so moronic; if it has 90% of the ‘best’ ones, doesn’t it make sense to download 1 archive than all individually? As long as Mr. Raible isn’t pawning all of ‘em off as his own, I don’t see a problem.
BTW, just ran across a company who’s essentially the same type of bundle (Eclipse3+truly free plugins).
July 24th, 2004 at 6:15 pm
He should have done the aggregation as an Eclipse plug-in. Then it could have included itself. Now that would have been cool!
July 25th, 2004 at 8:09 am
Javablogs is a great site, the only place where you can get some interesting stuff being worked on in the industry. Come on, cut some slack, its propriety :). Anyway, maybe javablogs.com should be open sourced
July 25th, 2004 at 11:14 am
Woo Hoo! We anonymous people did it! We took down the village idiot like a dog. He took his blog and went home.
Russell Beatte is no more.
http://www.russellbeattie.com/notebook/1007935.html
July 27th, 2004 at 6:08 am
Well what an interesting blog Rusell has now! I can see people queueing up to read it. Let’s see -
1)”let me point out a techincal fualt”
2)”The total comments count is confising”
3)”Thanks for the bug report”
4)”Iamgine leaving anaonymous comments on my site…what is wrong with someone ike that”
Oh yes, we’ll all be rushing to read this stuff over breakfast. What a stiff! Or should that be limp. BTW does anyone employ this asshat?
Biggus.
July 27th, 2004 at 9:55 am
Just checking out that old thing that as long as the first and last couple of letters of a word are correct, people can understand them :-)
Se wxat I medn?
July 28th, 2004 at 1:26 pm
You don’t like javablogs, send in a patch.
July 29th, 2004 at 7:33 am
Look out! JRoller is used on blogs.sun.com. Sun got infested with OSS.
July 31st, 2004 at 7:50 pm
care to show your list?
I mean, there is people (http://rc3.org/cgi-bin/less.pl?arg=6421) pointing out a list if supposed javites where half of the heads are smalltalkers,lispers,rubyists and pythoneers