On Thu, May 28, 2020 at 11:30 AM Béla Törös <[email protected]> wrote:
> > This seems to do a bunch of pre-calculation for things like min/max/avg > that are better done in PromQL. > > I think the main point is that you want higher resolution data, but > don't want to store individual results as they provide little data. > When we are trying to monitor certain things we want the results of > things like ping -c 30 every 30 seconds but definitely don't want to > store the results of individual echo responses. The problem with doing > things in PromQL with this type of data is that the resolution would > need to be superhigh to do meaningful comparison. > This is solved with things like Histograms. For example, this is how I implemented it in my smokeping_exporter. https://github.com/SuperQ/smokeping_prober > > > > ping_rtt_seconds{type=worst} > since last scrape, I would assume. > That's something you can't assume. Also, invalid way to handle it if you have HA Prometheus servers. > > > Doing this with `max_over_time()` in PromQL allows the user to decide > what timeframe to look at. > and by decide you mean any timeframe that is > scrape interval, right? > > We are definitely giving this a spin. Thanks for the contribution, > this looks promising to solve a few issues for us:) > > -- > B > > On Wed, 27 May 2020 at 20:54, Ben Kochie <[email protected]> wrote: > > > > This seems to do a bunch of pre-calculation for things like min/max/avg > that are better done in PromQL. This is both extraneous data, and unhelpful > data at the same time. > > > > For example, ping_rtt_seconds{type=worst}, worst when? What timeframe > does this calculate over? Doing this with `max_over_time()` in PromQL > allows the user to decide what timeframe to look at. > > > > Also, this is bad practice for labeling, because you have different > metrics for different meaning. > > > > For example: sum without (type) (ping_rtt_seconds) > > > > The results of this calculation are nonsensical. > > > > On Wed, May 27, 2020 at 10:32 PM Sebastian YEPES <[email protected]> > wrote: > >> > >> The Network Exporter is a mix of the blackbox-exporter and smokeping > with more specific features and metrics but specially with the addition of > the MTR module (traceroute). > >> > >> Source: https://github.com/syepes/network_exporter > >> Doc PR: https://github.com/prometheus/docs/pull/1647 > >> > >> > >> Have fun and enjoy your metrics! > >> > >> -- > >> 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/a47d97d6-94a4-4a6e-a9cc-60dcf21c8533%40googlegroups.com > . > > > > -- > > 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/CABbyFmoZE4t1LdgdtSFhrQ_gBfj7_LW6x_nNeG4w02OTL3Oa4Q%40mail.gmail.com > . > -- 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/CABbyFmoeeHx9M1skuk7iqFZsGk66nm5wJA%3D%3Dyr6P60r%3D5fia5g%40mail.gmail.com.
