I really wanted (needed?) Sentry at my last place but couldn't get approval for it.
Unfortunately we did already log all errors it was the (minimal, but existent) intelligence that Sentry did _after_ collecting logs that I needed. I wanted the histogram of errors and to get alerted when errors passed a threshold but also get alerted on any new error that hadn't been seen before.
Some of this can get hacked up with Grafana but you're really stretching it.
• It's huge. If I run `docker-compose ps`, it lists thirty entries, one of which seems to have failed and exited.
• With that hugeness comes CPU and RAM requirements. Sentry's installer complains if you give it anything under 8 GB of memory and 4 cores.
• The UI comes with an obnoxious amount of JavaScript, and is just generally slow and clunky. I see spinning wheels a lot.
Errorpush looks like a fantastic alternative for my needs, especially if I'm able to use the Rollbar Terraform provider to configure it, as well as if there are suitable integrations for all the programming languages my services are written in.
It is huge but with the pricing, spinning up a dedicated instance for Sentry is cheaper than saas version (if you already have an engineer to handle it)
What I don't like is it seems to randomly use certain amount of CPU despite very low volume of error reporting. No idea what it's doing.
It seems the one container is supposed to quit on launch?
I don't exactly feel the UI too slow and since it has features one would expect including user management against several auth providers, I can't find a decent alternative.
Their is also GlitchTip, which is an Open Source fork of Sentry, created after Sentry went proprietary in order to keep the Open Source legacy maintained. https://glitchtip.com/
I manage a self-hosted version at work for the development environment but we use the paid, cloud version for production environments.
Managing the self-hosted version is farily trivial (the catch is that, anyway, we don't provide SLAs or HA for the dev environment.
My main advice would be to:
1. use ZFS, keep the docker-compose files along with the data volumes on a dedicated dataset
2. take a snapshot before every update, delete the snapshot when users confirm everything is working okay
3. for the love of god make sure you perform every minor version.
I did have to do a 9.1.2 -> 21.5.x upgrade and it's been quite a PITA. In Sentry's defense, it was quite a jump. And in Sentry's honor, I was able to retain projects, users and most useful things (except events, but for the development environment who cares?).
Upgrading to, say 21.5 to 21.6 was quite a breeze instead.
It can be done, but it had weird moments of stuff breaking and being difficult to debug due to complexity, with only some dead forum threads as "help".
I really don't think it's a great idea to be "compatible" with a commercial product that has no link with the project: it can change at any time and break things.
This is cool - I've built a simple version of this sorta system before Rollbar et al existed and just stored the aggregated events in memory. I'd just keep the most recent occurrence only - not every event - and the times the exception occurred so I could see how often it was happening over the last N seconds (hours, days, etc).
Just wanted to say I think this is awesome and I love what you're doing here! :D
And to the author, I'd love to chat - drop me a line at cory@rollbar.com