On 09/06/2020 08:45, vivek kodira wrote:
Hi Folks,
This is my first question here and it is not specifically about
prometheus but about the way metrics are logged to a file. I hope you
can help.
We've implemented a service which logs metrics in the open metrics
format and as recommended in this page
<https://prometheus.io/docs/instrumenting/exposition_formats/#text-based-format>.
So typical metric entries will look like this:
|
app_disk_used_bytes 5.20941207552e+13 1588085025536
app_disk_free_bytes 7.1712562479104e+13 1588085025536
app_io_counters_read_bytes 2.4027136e+07 1588085025536
|
My doubt is - if/when errors occur when gathering a metric, how do we
log them in this file? Can the "value" be replaced with a text
indicating an "error". Or is the recommendation that errors not be
logged and be recorded elsewhere?
I asked this question on the git repo
<https://github.com/prometheus/docs/issues/1644> and was advised to
try asking here instead.
Have an additional metric that is a counter of errors, so if an error
occur that is incremented. You can then display or alert on the increase.
Another common option is an "up" metric (e.g. mysq_up) which has a value
of 0 if the metrics can't be fetched & 1 if all is OK.
For the broken metrics themselves you could leave them out (no metric is
returned at all during that scrape) or possibly return NaN.
--
Stuart Clark
--
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/64d73982-1264-0578-ec77-b754df8a7289%40Jahingo.com.
