Hacker News Reader

Comment by zimpenfish | Hacker News Reader

That's the reason I haven't bothered with Prometheus - pull systems make no sense to me since you have to configure a single place with perfect* knowledge of your system rather than just pushing local knowledge to a collector.

(Ok, you do need some knowledge at the parent if you want to raise alerts but you'd need that anyway.)

View on HN

You need a place with all that information anyway, otherwise how do you alert on something being missing?

If you can't easily do that with your existing infrastructure, you should fix that first. I've written about this at http://www.robustperception.io/you-look-good-have-you-lost-m...

If a stream of data stops appearing, then you can alert. You don't need to pre-configure the existence of that stream (although if you might never get data from an object, this obviously is a failure you won't catch.)

Your article is a good one but in my experience, many companies are still many years from being able to implement that kind of database:machine knowledge consistency.

You can only alert in that case if the data starts, and there's no transient issues preventing your monitoring working around the time it stops.

Alerting based on state changes is fragile, it's better to compare against what you expect to see.

You can leverage service discovery to make your configuration dynamic: http://prometheus.io/blog/2015/06/01/advanced-service-discov...

Reply on news.ycombinator.com