For me, the load balancer itself should be polled, and it as the sole accessor of the applications needs to expose metrics about them with proper labels
On Wed, Mar 18, 2020 at 11:29 AM Brian Candler <[email protected]> wrote: > On Wednesday, 18 March 2020 09:35:45 UTC, Jakub Jakubik wrote: >> >> then ignore or rewrite the instance label as it would be the same for >> both. >> >> > In that case you can actually return an "instance" label as part of the > scrape results, and set "honor_labels: true" in the scrape job to trust and > retain the instance label returned by the remote host. > > However, this seems to be completely the wrong thing to do here. You > should not be scraping two backends through a load balancer - you should be > scraping the two backends separately and directly. If there is no direct > network connectivity between the prometheus server and the two backends, > then enable it somehow, e.g. via port forwarding - or at worst use > PushProx <https://github.com/RobustPerception/PushProx> which can act as > a meet-in-the-middle. > > -- > You received this message because you are subscribed to the Google Groups > "Prometheus Users" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to [email protected]. > To view this discussion on the web visit > https://groups.google.com/d/msgid/prometheus-users/ce86f522-e4f0-4e81-9356-b3186ace1dac%40googlegroups.com > <https://groups.google.com/d/msgid/prometheus-users/ce86f522-e4f0-4e81-9356-b3186ace1dac%40googlegroups.com?utm_medium=email&utm_source=footer> > . > -- You received this message because you are subscribed to the Google Groups "Prometheus Users" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/prometheus-users/CALnV8WjyFSBTDBRcwNFcKOz3RpLvmxCUT%2Bpse2%3DMAYYSF88A%3Dg%40mail.gmail.com.
