Pushgateway metrics cleanup

[Aparna uday]

I could see API can be used to cleanup metrics based on job name. I would like to know if there is a way to cleanup metrics in pushgateway based on time (eg: clean metrics older than 2 weeks). 

[Julius Volz]

Hi,

There've been a number of issues and threads on this in the past, these might provide useful background info on the discussion:

- https://github.com/prometheus/pushgateway/issues/19

- https://github.com/prometheus/pushgateway/issues/117

- https://groups.google.com/d/msg/prometheus-developers/9IyUxRvhY7w/iUiIA5KLBgAJ

- https://groups.google.com/d/msg/prometheus-developers/QLOpK5iY2tc/p_qnMmoKBgAJ

Most of the responses so far have been: a TTL usually doesn't solve a problem that would come up in a use case that the Pushgateway is designed to handle.

Regards,

Julius

On Tue, Jul 28, 2020 at 11:45 AM Aparna uday <apr....@gmail.com> wrote:

I could see API can be used to cleanup metrics based on job name. I would like to know if there is a way to cleanup metrics in pushgateway based on time (eg: clean metrics older than 2 weeks). 

--
You received this message because you are subscribed to the Google Groups "Prometheus Developers" group.
To unsubscribe from this group and stop receiving emails from it, send an email to prometheus-devel...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/prometheus-developers/5da59638-f9d4-4c76-9df3-0b793b6656fao%40googlegroups.com.

--

Julius Volz

PromLabs - promlabs.com

[Braden Schaeffer]

I have yet to understand how push gateway is both simultaneously designed for handling batch job metrics but is also not a great use case for batch job metrics..