Zeebe Incident Reporter

Josh Wulf: Here is a tutorial on writing an exporter: https://zeebe.io/blog/2019/05/exporter-part-1/

Josh Wulf: Filtering (like on Incidents) is explained here: https://zeebe.io/blog/2019/05/exporter-part-2/

Dan Shapir: Thanks! I’ll create such an exporter. What do you mean merged to core? How?

And I’ll check your PR didn’t see it for some reason!

Josh Wulf: I started it as a POC (in Kotlin) here: https://github.com/jwulf/zeebe-incident-alerter

Josh Wulf: Actually, I can hear <@U6WDGBHT4> in my head

Josh Wulf: I built this, but here is the problem

Josh Wulf: If the webhook server is not available, you have to either lose the alert

Josh Wulf: or make a choice about a retry strategy - at which point you start using broker resources

Josh Wulf: And ultimately you just defer the inevitable - at some point you say “we abandon retries” or else you say: “we keep retrying all the buffered incidents until the server grinds to a halt”

Josh Wulf: Neither is a good outcome

Josh Wulf: Whereas if you listen to the Elastic Search export, you can trigger from that

Josh Wulf: Exporters run on a single thread, so blocking in there is an issue

Josh Wulf: Using async operations involves creating your own thread pool, so you are messing with the core broker’s resources

Josh Wulf: Anyways, a working POC now available here: https://github.com/jwulf/zeebe-incident-alerter

Dan Shapir: Thanks!! I actually think that if it’s down losing the alert is legit…

Josh Wulf: In that case, it’s just what you ordered :rolling_on_the_floor_laughing:

Dan Shapir: <@UTM6C2C3H> Zeebe Incident Reporter

1 Like