Ben Kochie
Ben Kochie
This is probably not a good feature to add. If you have apps that you want to support Prometheus, add a direct Prometheus listener endpoint.
I'm thinking it should get the end timestamp of the delta accumulation.
Yes, my thinking is that because there is some jitter in the timing of each real-time poll, and when Prometheus performs the scrape.
I'm going to continue to do some debugging on my end. I'll get back to you on what I find out.
Since it's basically polling another monitoring system, yes Another way we could handle this, is to convert the exporter to become a remote write streaming sever. Rather than scrape ->...
We tried to upgrade from 3.0.x to 3.3.x and it looks like we're seeing something similar. The process starts up and bursts to 1500Mi memory for a few minutes, then...
:tada: @bwplotka
Looks like this needs a rebase
Adding a metric seems like a good idea to me.
Maybe we should move some of the text to other documentation files?