Refresh timeout for quantile metrics

[skul87]

Hello, @martin-helmich

prometheus-nginxlog-exporter version: 1.10.0
input - syslog udp.

I have POST-request which executs for 28.749 seconds.
Requests are rare, about 1-2 request in 3-5 mins.

22.22.22.22 - - [08/Dec/2022:17:57:54 +0300] example.com "POST /provider/balances/url HTTP/1.1" 499 0 "https://example.com/provider/balances/" "Mozilla/5.0 (Windows NT 6.3; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/102.0.0.0 Safari/537.36" 31006 650035276 28.749 1670511474.786 - 28.749 - 194fa74989685fade69f96b2eecdfc09 -

Seems like nginxlog-exporter caches quantile metric, so Prometheus gets same value multiple times(about 10 min).
Prometheus scrape time 30 sec.

PromQL request
nginxlog_http_response_time_seconds{hostname="monolit-balancer-1",quantile="0.99",method="POST",status="499"}[70m]

response

nginxlog_http_response_time_seconds{app="main-shop", hostname="monolit-balancer-1", instance="metrics:4040", job="nginx", method="POST", namespace="balancers", quantile="0.99", status="499"}
...
0.163 @1670511288.077
0.163 @1670511318.077
0.012 @1670511348.077
0.012 @1670511378.077
0.012 @1670511408.077
0.012 @1670511438.077
NaN @1670511468.077
28.749 @1670511498.077
28.749 @1670511528.077
28.749 @1670511558.077
28.749 @1670511588.077
28.749 @1670511618.077
28.749 @1670511648.077
28.749 @1670511678.077
28.749 @1670511708.077
28.749 @1670511738.077
28.749 @1670511768.077
28.749 @1670511798.077
28.749 @1670511828.077
28.749 @1670511858.077
28.749 @1670511888.077
28.749 @1670511918.077
28.749 @1670511948.077
28.749 @1670511978.077
28.749 @1670512008.077
28.749 @1670512038.077
NaN @1670512068.077
NaN @1670512098.077
NaN @1670512128.077
NaN @1670512158.077
...

Maybe I have missing something or doing bad logic, but i think here need add timeout for refresh data?

[skul87]

I meant something like this:
max_idle_duration: 1m
https://grafana.com/docs/loki/latest/clients/promtail/stages/metrics/