Ask HN: What exception tracking service do you use? | Hacker News Reader

Ask HN: What exception tracking service do you use?

30 pointsflippyhead43 comments

There's quite a few including BugSnag, AirBrake, HoneyBadger and more I'm sure. Any recommendations on which is best?

Not sure if this surprises anyone but it surprised me when I found it - I use Google Analytics for iOS. Their library handles platform exceptions and I use it as a catch-all for ones I didn't see coming: https://developers.google.com/analytics/devguides/collection...

I also use Test Flight's reporting to do the same but I don't push that lib to production usually, where as the GA one is already handling analytics, so it ships.

Email with stack trace attached, and md5(filename + line number) as the thread ID so that errors from the same place get grouped together.

Me too on this. I feel like someday something will outshine this, that day is not yet today. (for a small/medium python web app)

We use Bugsnag. It's been great and easily integrates into a variety of platforms.

James from Bugsnag here, we're constantly rolling out tons of new features to speed up the find/fix/deploy bug hunting cycle. I'm surprised so many people are still using emails for error monitoring, using a hosted service like Bugsnag/airbrake/rollbar will massively help with productivity and actionability.

James I didn't realise Bugsnag was you - in that case we'll give it a roll. I've been wanting something to switch to

Even though it's not its highlight feature, New Relic. Its panoply of other amazing features make it well worth it.

New Relic is a god sent.

Have you looked at AppNeta?

Self hosted sentry, very pleased with it.

Personally, for a Django project, Sentry is probably the best integrated solution.

We use the self hosted solution. Works great!

We're using HoneyBadger at Magoosh. It's been working quite well for us. The Capistrano integration for tracking deployments is really handy.

We used to use Exceptional because it was really cheap, but maintenance and bugfixes seemed to have stopped so we made the switch

+1 for Honeybadger. Moved away from Airbrake and we've been happy campers.

Using Sentry here at FlightCar. Have had very good experiences:

https://twitter.com/cjbprime/status/347467911050448896

Disclaimer: I work on Exceptional and Airbrake.

We eat our down dog food and use newrelic and logentries. We've a few big updates in the coming months. If you would like to get early access please email ben@airbrake.io

Don't know which is best but we used airbrake and getexceptional in the past and it had massive issues. We now use BugSnag on 4 rails apps (ruby and js) + one node.js app and it just works.

What kind of issues?

We stopped getting exceptions a dozen times from a few hours to a couple of days. And no the exceptions didn't came up after that delay, it's like they were never sent.

After using an email-based homegrown solution that I thought was sufficient, I gave RayGun.io a shot. I use it for .NET exception tracking and I am about to wire it up to a Rails app, and a very very large Backbone JS application.

It is really quite good (at least for .NET). The management UI is great and does a very nice job of grouping duplicates, similar issues, etc.

I already use some dev tools from Mindscape (their .NET ORM, LightSpeed) and their products and support are top notch in my experience.

http://raygun.io

Rollbar here as well and TestFlight for iOS. We'd love to use Rollbar for iOS as well, but the support isn't great right now. That said, their support team has been great.

For non-web, home-made exception handler that dumps the trace to syslog. Then, FATAL and ERROR-levelled entries are forwarded by email (rsyslog's ommail module; http://www.rsyslog.com/doc/ommail.html).

For web, I mostly use Django and Flask and both have [almost, in Flask's case] built-in one features, that sends an email with backtrace and request details when a bug happens.

Should I pay more attention to alternatives?

Sentry. We pay for them to host it.

+1 - A great example of open source also generating a good business - for users and hopefully the author :-)

For the .NET crowd, ELMAH (https://code.google.com/p/elmah/) is the way to go.

Recently attended a presentation on Appfail (for .NET) by its founders. Liked the product and team: http://appfail.net

While not exception tracking per se, we use New Relic and we also log exceptions to a special Graylog2 stream. It's working very well actually.

Rollbar. We used to use Airbrake but Rollbar has been much better for us. It is still pretty new but they are constantly adding features.

We've been using Airbrake also, have been experiencing quite some issues with exceptions not showing up and slowness of the web ui.

I've been using Sentry [1] lately and it has been excellent. Only issue has been that using SSL from Java was quite troublesome as their certificate is not supported by the JVM by default [2].

[1] http://getsentry.com [2] https://github.com/getsentry/sentry/issues/903

Edit: Rollbar looks interesting. Too bad they don't have an API for Java.

I have used the Rollbar Clojure client library and I found it to be adequate FWIW

I've been really disappointed by how little development has been put into Airbrake. It's hardly changed or improved in the all the years I've been using it. I'll look forward to checking out Rollbar.

Sentry on heroku, https://github.com/dmishe/sentry-on-heroku or any fork

ProjectLocker (where I work) has integrated exception tracking, so I use that. Especially useful: Twitter notifications of exceptions, which show up on my phone.

https://gist.github.com/bobvawter/5986508

Emailed stacktraces to my personal email :D

We use airbrake with self-hosted errbit (opensource airbrake api-compatible clone) server.

Both them and exceptional seem to be in zombie mode. Massive bugs that have remained in place for what feels like years, no new features etc. I'd like to go with someone who cares!

That's what errbit is for. You only have to use the Airbrake libraries for reporting the errors, and errbit hosted locally collects/notifies/displays them.

If you already integrated with Airbrake, you just have to reconfigure the host it reports to, to your own errbit server and it'll swap everything over from deploy tracking to error reporting.

Errbit is an active project, https://github.com/errbit/errbit, and error reporting libraries don't tend to need a lot of maintenance.

Honeybadger so far. Working well enough. Love the deduping.

What is an "exception tracking service"?

When your code crashes, it (usually) generates a report telling you what went wrong. This is an exception report.

An exception tracking service provides an place that you can send, store, and index these reports. It'll usually allow you to view them, search them, group similar errors, and generally provide tools that are useful when trying to resolve the problems that are occurring.

Thank you. Soothing like the Windows event viewer ...

Errbit

Reply on news.ycombinator.com