Hello, absent_over_time() Looks good. It should allow me to alert If metrics is Not available for 10min. This will not alert If system reboots.
If i use a query with Offset I think it would alert only AeS Long as the Offset still hast a value, eg Offset 10m i will get an Alarm For 10m but after that the "now" is the Same as the Offset. Will give it a try. Thanks!! Chris Siebenmann schrieb am Montag, 26. Februar 2024 um 17:22:10 UTC+1: > > Will I run into issues with "staleness" if there aren't any metrics > anymore > > for (more) than 5 minutes? > > Or perhaps can I use this "staleness" indicator in some way? > > Perhaps this is a use for absent() or absent_over_time(), if you know > specific metrics that should always be present from the push sources > (and you know the push sources). > > It might be possible to craft something clever with 'offset' and > 'unless' to filter out metrics that are still present, eg: > > pushed_metric offset 10m unless pushed_metric > > I think this will give you every pushed_metric series that was present > ten minutes ago and isn't now (because it's stale, since it hasn't been > pushed recently enough). This is less clear than an explicit absent(), > but means you don't have to statically know the job/instance/etc labels > for all push sources that should be there. > > - cks > -- 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 prometheus-users+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/prometheus-users/e659e395-8256-43f4-9031-d369dd8bc7efn%40googlegroups.com.
