Real-time crash reporting tool Sentry raises $9M | Hacker News Reader

Real-time crash reporting tool Sentry raises $9M

techcrunch.com

214 pointstermostaatti67 comments

sigh - I really love sentry to bits. Very responsive support, all open source code, nice user interface, no nagging for additional services and very fair price (to the point where it's much cheaper to just host with them rather than trying to self-host, despite the thing being extremely well documented and fully supported for self-hosting).

I wonder what we'll lose first before they will eventually be sold to get the money back for the investors :(

Yes. I'm cynical, but it's always the same path in which good products eventually are shut down or get way too expensive.

On the other hand, I would be extremely happy if it was Sentry to prove to the world that you can in-fact take $9M in investment as an Open Source project and still stay independent, keep your original values and not be shut down by an eventual buyer.

Hi Pilif, I led the Seed round in Sentry and am on the Board representing Accel.

The single most important part of investing in Sentry for me was the fact that it's Open Source and people love using it. The funding is intended to continue that tradition while also allowing Sentry to more aggressively invest in better mobile support, better support for other languages/environments like C# and SpringBoot, a deeper feature set (Breadcrumbs is a good example http://blog.getsentry.com/2016/05/04/breadcrumbs.html), better support, etc.

Once upon a time I wrote software and I was always annoyed that I couldn't only use Open Source tools so I could audit their code and fork/patch them more easily as needed. Part of the reason for that was that investors have often shied away from products that cater to developers dismissing them as bad business. I think that's bullshit. If everything is going to become software, then the people building software are the most important folks in the world. People like you willing to pay a fair price mean it's quite possible to build a huge business.

It's important for us to continue to build a great, Open Source product and offer a SaaS option that's cheaper than doing it yourself. We plan to continue doing that.

For now, the thing I don't understand is all the people using proprietary crash reporting tools. Our goal is to make Sentry (Open Source or Saas) the strictly dominant solution for crash reporting. If we can accomplish that and offer a fair service, I think we'll have a good enough business for me :)

> allowing Sentry to more aggressively invest in better mobile support

Why would an operations tool need mobile support? Are we using mobile phones for systems administration now?

I think he's referring to better support for crash reporting inside of mobile apps.

pfg is correct

David here, founder, CEO, and extremely stubborn with my perspective. To echo what Armin and others are saying, we're going to prove that you can build a company around an open source product without turning it into crippleware. There will be things that end up closed source (there already are, like our billing code), but Sentry itself won't be one of them. We're in a fortunate position in that we've been an established (albeit, small) business for a number of years now, so we're able to build off of that existing success and good will.

Great answer David. At GitLab we self host Sentry and love the product. I understand that pilif has reservations, some open source projects prioritized profit over stewardship after raising money. What we did at GitLab is write down how we think about this in https://about.gitlab.com/about/#stewardship You may or may not want to do something like that at some point. Anyway, lots of success with Sentry, the product is great and the upcoming features sound awesome.

Huge props to GitLab. Definitely ahead of us organizationally, but one of the few companies that I pay attention to in the space that seems to be similarly aligned.

Thanks for the kind words Zeeg. Please let me know anytime we can help with anything.

is there anything structural about the deal that would allow this?

even bill gates was ousted from microsoft as chairman[1]. Of course, by then he held only 4.5% of the $277B company, so he was only its single largest shareholder and only held about $12 billion in Microsoft stock, a pittance.

if anyone is serious about stubbornly sticking with their perspective, they need to have the structure in place to do so, in my opinion. regardless, good luck going forward: it's bound to be a good outcome for you.

[1] http://www.idigitaltimes.com/bill-gates-fired-microsoft-inve...

I have been using Sentry since Cramer showed it to me for del.lic.ous way back in the day, and I can concur that he means what he says. Now if only he'd remove that stubbornness towards Node.js... :P

As long as it remains open source, nothing prevents you from self-hosting, tho

You first try self-hosting but when things get serious, you usually switch to SaaS version of Sentry because maintaining an extra server and an application which you don't familiar with is not that easy. Open-source is actually a marketing strategy here. It makes the barrier for you to try the software is quite low which is extremely important for both startups and enterprise companies. Even some enterprise companies don't want to pay when they're evaluating your product.

If you don't want to pay, then host Sentry on your server, the company behind Sentry doesn't lose anything. In fact they can still benefit from you since you became a part of the community. These people usually advocate Sentry, report bugs etc.

I mean, thanks for describing how many open source companies make money?

It's relatively new in the industry to make money from open-source projects. There are enterprise companies such as Redhat but startups usually scared open-source because of your concerns. I admit that Sentry's approach is actually is bit risky because Sentry is easy to install and maintaining is not that hard unlike the other open-source startups that develop databases.

However you need to differentiate your product from your competitors and error-reporting is not actually quite a hard problem, I know many companies that still uses e-mails (send email with stack-trace directly from the app) for error-reporting. Sentry has a unique approach here, it's an advanced error-reporting tool and tries to build a community around Sentry with open-source.

If you're a developer who needs error-reporting for your new fancy app, instead of your custom solutions (which are actually workarounds) install Sentry easily and get an advanced error-reporting application for free. Eventually when you work in a enterprise or start a company, you will want to stick with Sentry since you're familiar with it and don't want to take the responsibility to host Sentry so you will switch to SaaS solution.

I'm not actually related with company behind Sentry, these are just my thoughts. :)

Which is why that's probably the first thing to go away.

Definitely not. We're an open source project and have no desires to change that.

Your investors might have desires to change that though.

I have been a participant to these discussions often enough so I'm quite aware of public perception on those topics :) I think we should be judged by our performance though and we are doing our very best to make sure that people like the product.

It would be crazy business to throw our biggest competitive advantage over board. It's not that we do not know why people use us :)

Honestly, I think you have the best UX and the best support team and the most compelling price. I just hope this remains the case in the future and I wish you all the best of luck.

I will be incredibly happy to be proven wrong on my fears.

I trust Armin given his excellent open source track record in the Python community. I will be using Sentry when I need to, solely because he works there :)

You're overstating how much power investors have at this stage of a company. During a seed, Series A, Series B etc. the entrepreneurs are often still in complete control and investors are mere advisors.

It is the companies that raise crazy amounts of capital over many rounds who end up at the whim of boards and investors.

Equity isn't where all the control comes from. Because they raised 9mil it's almost a sure thing they'll have to raise another round or go bust, so there's huge incentive to do what investors want.

We do not :)

Also, hi Nik, been awhile!

We've self-hosted Sentry for years... it's pretty effortless. I'm a huge fan.

Most products are not open source to start with, and are much more profit-oriented from the start.

Reminds me of old question: what a socialist says when you give one some gratuitous money? "can I have some more?" (I'm not saying that this is related to socialism)

Right, the standard procedure is to carve up the app into the open-source piece, and then the "enterprise" / "commercial" pieces which will be closed source.

Nah, I think the most common procedure still is to develop propiertary closed-source software, and charge for that. Sentry is the exception here.

$10 says the guy in the Christmas sweater is Ops/DevOps (I'm DevOps myself). :-) lol

The length of the beard is proportional to ops experience.

What's your PayPal address?

Am I wrong?

Nope. That's me. :)

You're in ops? I could have sworn I've seen your name all over raven-js.

Funny, what is it about DevOps that either attracts or grows beards?

The lack of time to shave from continuously-deploying the code to make the alert dashboard stop blinking yellow.

It's a well known phenomenon called the "UNIX beard" [0]

[0] "The length, bushiness, and unkemptness of the Unix beard are all directly proportional to the owner's expertise." Source: urban dictionary

> The length of the beard is proportional to ops experience.

I know you're trying to be funny, but this statement excludes all people who can't grow long beards from being considered "ops".

Why did you have to turn this into a sexist argument? OP was clearly just joking around and not trying to suppress non men

Lots of men can't grow a beard! Reading sexism is GPs comment says more about you than about them.

As someone with a lot of indio blood, I can't grow a beard. But I did grow a sense of humor. Kindly fuck off.

> Kindly fuck off.

After you

Also, not everybody has such a cool Christmas sweater!

Or does it? http://i.imgur.com/goYgkVd.jpg

I love sentry and i like mitsuhiko projects: flask, jinja2 and werkzeug. Sentry substantially improved error handling and realiability in my startup

Congratulations guys! When I read that CCP Games was using Sentry on the PS3 for their multiplayer game Dust 514, I was pretty damn impressed.

Armin working there is what sealed the deal though - I've been following his work since I discovered Flask a few years back.

Long story short, I will be using Sentry for my next web app.

I had the pleasure of meeting Cramer and his team last year at their office and left convinced these guys are very talented. I look forward to seeing what they can do with $9 mill.

I think the biggest problem with Sentry is that it will sample exceptions (I wonder if there's a way to switch this off with $$$?), meaning that I can't rely on it to view and examine every instance. Sometimes you need to fix things manually which requires every exception to be saved.

Since I want to get every single exception logged, Sentry feels quite useless and I wouldn't select it for my next project.

Other than that it has been a fantastic tool.

We're working on improving this. It's a tradeoff we made early on as our primary customers have always been very large tech companies (meaning lots of errors). In those cases individual events have little to no meaning, and what you really care about is the aggregate view.

That's very different than if e.g. you're a payments provider, and you literally need to know every transaction that erred, but that's also quite a different use case. In my opinion, you'd want to guarantee those errors are logged elsewhere (traditional audit logs), and utilize Sentry primarily for the workflow and alerting features.

But if I need another tool for logging those errors, I could use the same tool for both use cases, which would make Sentry unnecessary?

Ideally speaking a Sentry exception is the last escape hatch - if we can't handle something in code at least the error is sent somewhere so that developers can handle the issue. Following this the philosophy I'd like to follow is that every time I get a Sentry exception, I'll make a new ticket in Jira "handle error x, see link to Sentry", and ultimately there wouldn't be any errors sent to Sentry.

FWIW my company uses both Splunk and Sentry, I've found myself using them to combo together quite nicely

If you run your own instance, you can turn off sampling: https://docs.getsentry.com/on-premise/server/config/#data-sa...

I've built a mobile crash-reporting solution around self-hosted Sentry. (To my chagrin, I've been unable to open-source the bits I built around Sentry.) Sentry is a great product and David et al are a great team. I'm hoping them continued success.

You should just come work at Sentry!

Been using Sentry since early versions (even contributed some crappy plugins myself) and moved to hosted service immediately. Great tool, I am glad to hear this news!

Hah! I used to own this domain, planning to use it as a tool for uptime monitoring and statistics. I'm glad that someone else is making real use of it!

Great news! We've been using hosted Sentry for a while (couple of years) and it works really well for its price. Highly recommended. P.S Has anyone tried Golang support yet?

Congrats! How do you all feel this compares to TrackJS?

TrackJS is only JavaScript, whereas Sentry is cross-platform. In many cases a niche single language service works very well, but specifically for Browser JavaScript we go above and beyond most implementations and it's one of our strongest offerings.

Good for them!

On a side note, GCE has a similar feature with Stackdriver: https://cloud.google.com/error-reporting/docs/viewing

It's not really the same thing. Sentry is far more general purpose, and doesn't require GCE. If your language or framework lacks support, you can add it. Also has a bunch more functionality.

Google's offering is great on App Engine (I've used it a good bit), but that's the sweet spot.

Congrats David and Team!

How is this compared to Apteligent/Crittercism?

The best way to understand the differences is to try it out. I haven't used either of these tools personally, so my comments are more general.

We focus on giving you an extremely rich crash report and day-to-day workflow tools, whereas often other tools are more about high-level passive analytics. Additionally we work cross platform (web, mobile, and desktop), and go much deeper than most things that simply aggregate logs.

I'm the co-founder of Apteligent. David, I'm surprised you would describe our solution as "high level passive analytics" unless you haven't tried it yourself.

I would say the difference is our deep focus on mobile user experience vs Sentry's focus on cross platform error reporting. To understand your users this also means you need more data than just crash. For example: app load times, network data, UI latency, etc.

Apologies (edited), I meant to suggest that other solutions are generally that. I definitely haven't used Apteligent (or Crittercism), but things like New Relic (which from a glimpse of screenshots, seemed similar), I would label as passive systems.

congrats guys

Thanks Nik!

Reply on news.ycombinator.com