Brad’s Musings

Way to present a one sided argument! I love your use of childish language to portray how advanced your critique is. Can't wait for more.

I wish more people would spend time spewing hate at open source software than actually contributing their time in a positive way to making it better. How much did you pay Tom to try God again? What do God support licenses cost? Did you pay for one?

Really people, open source software is out there to improve your life. There are never any guarantees that it will work and in return you never have to pay a dime. I wish more people would understand that trade.

I have to say, we use God and it works (and I love open source) and I realize it's free software and it's hard to write good software...

But yeah, not really stopping process on stop, and the crazy memory leaks (why would a monitoring utility take 500meg of memory after a week or two of usage?) mean Brad is just speaking the truth.

Brad, hook us up with some monit start/stop/monitor scripts. ;-)

@cardmagic http://en.wikipedia.org/wiki/Ad_hominim

Garry: I never claimed he was lying, just that he presented a one sided, and very short-sighted critique that did more harm than good to anyone reading it.

johndouthat: I think you have to re-read that wikipedia page yourself. Not once did I attack Brad, I only talked about this article.

@cardmagic you're missing the point. I don't care if a person releases a sub-par OSS project and the community slowly makes it better. That's called progress.

I have a problem with people who are not forthcoming about the shortfalls of their projects and end up wasting a lot of peoples' time. Tom presents a one-sided picture of how "awesome" God is when it has some serious flaws. To be more productive and advance God, he could prominently list these flaws on the God website and look to the community for answers.

Wow, can't even spell my name right! It's "Tom Preston-Werner", bro. This might get my blood pressure up a bit if it didn't read like it was written by a child.

Honestly, I wrote god one afternoon after monit made me angry enough to take a stab at process monitoring myself. Eventually I worked on it a lot at Powerset where it was used successfully on every production server. We still use it in production on GitHub and it performs extremely well. God allows me to write custom conditions and operations that monit doesn't offer.

Sure, on some architectures it leaks pretty badly, and I do blame MRI for that. I've eliminated a good deal of leakiness, which you can read about at http://groups.google.com/group/god-rb/browse_thread/thread/01cca2b7c4a581c2. I'd love to figure out the rest of the leaks, but I'm less motivated to do so when it doesn't leak on Engine Yard's install of MRI.

I'm sorry god doesn't work for your situation, but you are responsible for evaluating software and managing your own time, not me. Enjoy monit!

@mojombo Sorry about that. Fixed! You're right about evaluating software, but OSS developers can do very simple things to make that process easier.

The "awesome" claims are what stand out in my head and get my blood boiling when God can't even restart a process. Then I think about all of the other people who read the god website, believe it word-for-word (it even made it into Railscasts), and waste their time.

I couldn't find anything that documented all of the bad aspects of God so I felt compelled to write this post. I've already received a few "thanks" from people who were thinking about using God.

Funny, there's a prominent link to the Google Group at the very top of the web page wherein people expound in depth upon the shortcomings of god. Some of them even offer to help fix the issues instead of writing nasty articles about the author. http://groups.google.com/group/god-rb

It really seems like you've been caught cutting corners on evaluating software and now you want someone to blame.

@KirinDave it took me 6 months to evaluate God and you're reading my findings.

I'm reading your findings and frustration. But you didn't bother to check the google groups first? The point I'm trying to make is that documentation and discussion of God's shortcomings is not hard to find and has not been hard to find since @mojombo did his big ruby memory leak post last year.

@KirinDave of course I read the Google Groups, that's where I confirmed the problem with processes not stopping. The frustrations are out of an overall problem of OSS projects self-promoting themselves with unsubstianted, border-line misleading claims like, "God - (like monit, only awesome)".

Admittedly, I'm Brad's co-founder so I'm biased. But -

@cardmagic - "Really people, open source software is out there to improve your life. There are never any guarantees that it will work and in return you never have to pay a dime. I wish more people would understand that trade."

I believe Brad's point is not against OSS - in fact, isn't he just advocating a different OSS solution? (Albeit in his typical flagrant tone, which is not as reasoned and measured as yours and mine, fair sir).

Bashing most OSS is like making fun of a charity, it is in very poor taste since the intentions behind them are for bettering people's lives

I'm amazed at how obtuse @cardmagic is about my experience with God (the process monitor). He still thinks I'm "bashing OSS" which is not the case; I'm bashing crappy software that is advertised as otherwise. Reread my post a few times and get past your knee-jerk reaction to somebody criticizing the marketing tactics of an OSS project.

What a coincidence! We just decided to stop using God.rb last week because it was hogging over 50% of our memory. Another issue I had was the unwatch command yielded strange issues:

Running two daemonized processes: orbited and activemq, I would unwatch orbited, check status and it was still watching activemq which was good. Then I would unwatch activemq, check the status, and orbited would be watched again! Wack.

I'm moving over to monit as well, I'm going to go read your latest post. =)

look, both god and monit have sucky points. i've been using them in production deploys with hundreds of service instances for as long as both have been in existence, minus a few months in each case.

cardmagic (lucas carlson), when was the last time you fixed a memory leak and submitted a patch for MRI? yeah, thought so. you've been making a living off of ruby for how many years since you gave up php scripting now... 3? isn't it about time you spent some time on the core implementation of it?

the claim was that god is awesome. that's really typical hype. typical to ruby developers who re-invent a wheel and think they've done everyone a favor, when all they've done is built a leaky boat using a crappy implementation of an otherwise nifty language.

so yeah, call a turd a turd. yup, god's a bit turdlike. but so it monit. but at least monit's authors aren't telling you its awesome, and it isn't written in something that leaks.

use the right tool for the job. it isn't always going to be written in your language of choice.

@maje I have not written a blog post calling Ruby a turd for having memory leaks, and I never would because I would spend time working on a patch instead if it upset me that much. After all, I never paid Matz to fix those bugs for me.

@cardmagic This article has saved me time. I am grateful to Brad for writing this article. Monit is a sharp tool written in C. It would have been mind-boggling to using a process monitor that leaked memory in lieu of Monit. I would struggle to understand what was so "awesome" about that.

Anyone who engages in this sort of hyperbole, calling their project God, making breathless claims that they are like a stable and proven software project, like Monit, except that they are awesome, had better deliver. If you say all that in order to get people to install a runaway Ruby script as daemon on their servers, they are going to come back and use language calibrated to deflate the hyperbole.

Open source software is business. It is not charity. This God of yours is not putting food in people's mouth. Open source is a licensing scheme. No one needs to show gratitude to a person, simply for attaching an OSI compliant license to their digital squiggles. No one here is an ungrateful benefactor.

Open source is a strategy to grow software, to encourage adoption, and collaboration. But it is just one strategy. Another strategy is to be honest about what you are marketing, still another is to provide an excellent basis for further development. Calling the software God, awesome, better than its competitors, is either dishonest or deluded. Either way, it doesn't give the impression that there is someone there willing to listen. It's all bad business.

Who here is attacking Ruby? I am grateful to Matz. I use his wildly successful language. It makes my life easier. It helps me create new things. You are not entitled to the same sort of fidelity that I show to Matz because I had the misfortune of cloning your repository on GitHub. That is not how open source works.

Brad is warning people that this project is going to waste their precious time, which is more important to me that stroking the egos of people who call their software God, claim it's awesome, and then get their panties in a bunch when someone says that, the claims are "bullshit" and the software is a "turd".

Maybe, if the original authors had not made such wildly exaggerated claims, then Brad would have said it is was unpolished, maybe I'd be installing something not called God, but given the claims to divinity, I am happy that Brad has been so blunt in his description of the software. Now I can install Monit and rest assured that it is the best tool for the job.

I'm in the process of reworking my production stack (currently using monit), and came across this article, which confirms my suspicion that using a leaky ruby process to monitor a bunch of other leaky ruby processes is perhaps not the best idea out there ("Quis custodiet ipsos custodes?", anyone?). I'll stick with monit, which in my deployment environment is consistently one of the longest-running processes, and maintains a modest memory footprint. I don't know what Tom's original frustration with monit was, that caused him to want to reinvent that particular wheel, but for my modest needs, after initial configuration, monit has been nothing but smooth sailing.

Stop bitching and start fixing.

No, Sam. There is no need to start fixing. Monit works just fine.

The purpose of this article is to warn people that someone is trying waste our valuable time by misrepresenting their software. Well done Brad, and thank you once again. Monit is awesome.

I, for one, appreciate the article. After running god two separate times with terrible results (leaking to a gig of memory within a month early in the project's life - monitoring software shouldn't need to be monitored - and more recently having it simply go quietly AWOL. Would not restart and would give no errors). I had run monit for years without a single incident (but liked that god was in Ruby and later that it had complex conditions thus the switch). It convinced me to kill my 2 day spike of trying to get it to work correctly and go back to my tried and true monit.

While sure the wording of this article was strong and at times childish, the claims of god since day one have been horridly hyperbolic and in my experience no way connected to the actual state of the software. Terrible documentation, bad feedback when something goes wrong. Like one person said, calling something that's basically in alpha, or in beta at best "Like [established project in it's 5th full version] but awesome" with no preconditions, or disclaimer is childish and misleading. Sure no one is paying the author but that doesn't mean they aren't costing me money. It's a trust thing, the developer makes a claim, maybe the community backs it up, and I make a call about investing my time and credibility with my peers into it.

In the case of god it was a bit like the Emperor's New Clothes. No one had said a bad thing about it or if so it was quiet or early on (google groups don't count to me as people are there because of problems - if we go by groups and irc channels every project sucks). After a couple years of existence and the documentation and product is barely better but the claims are still just as large I think a call out is due. It's about managing expectations.

And to those who do the "if you don't like it commit a patch" dance: First, people seldom have time to contribute patches to things just to get them to a base level of working, esp when there are established players out there. Usually it's about "Man this is great, if only it did X" (which is where I am at with monit...again). Second, OSS projects are cults of personality. Tom's response here alone is enough reason for me to not want to *use* the project again much less contribute. It sadly taints my otherwise very high opinion of the guy and other stuff he's worked on. Sometimes silence is the best responses to critiques. Esp if you find them particularly childish.

Tom - Could you say more about EngineYard's MRI install?