Thanks for the hint. I checked the Go DNS feature and found these hints:
1. export GODEBUG=netdns=go # force pure Go resolver 2. export GODEBUG=netdns=cgo # force cgo resolver I tried to set the cgo env variable and restarted services. however systemd-resolved and nscd seem not to be able to cache it. May have to wait for a colleague who is more experienced in linux than me. perhaps we can figure it out why it's not working with the new behaviour. Ben Kochie schrieb am Freitag, 15. März 2024 um 17:52:09 UTC+1: > All of the Prometheus components you're talking about are > statically compiled Go binaries. These use Go's native DNS resolution. It > does not use glibc. So maybe looking for solutions related to Golang and > nscd would help. I've not looked into this myself. > > But on the subject of node local DNS caches. I can highly > recommend CoreDNS's cache plugin[0]. It even has built-in Prometheus > support so you can find how good your cache is working. The CoreDNS cache > specifically supports prefetching, which is important for making sure > there's no gap or latency in updating the cache when the TTL is close to > expiring. > > [0]: https://coredns.io/plugins/cache/ > [1]: https://coredns.io/plugins/metrics/ > > On Fri, Mar 15, 2024 at 3:41 PM Alexander Wilke <nachtf...@gmail.com> > wrote: > >> Hello, >> >> I am running blackbox_exporter and smokeping_prober on a RHEL8 >> environment. Unfortunately with our configu wie have around 4-5 million DNS >> queries per 24hrs. >> >> The reason for that is that we do very frequent tcp queries to various >> destinations which results in many DNS requests. >> >> To reduce the DNS load on the DNS server we tried to implement "nscd" as >> a DNS cache. >> >> However running strace we notice that the blackbox_exporter is checking >> resolve.con, then nsswitch.conf then /etc/hosts and then send the query >> directly to the DNS server not using the DNS cache. Thats for every target >> of blackbox_exporter. >> >> For smokeping_prober I am aware that it resolves DNS only at restart and >> we notice the same. All requests are directly send to DNS server but not to >> the cache. >> >> anyone using nscd on RHEL8 to cache blackbox_exporter and/or >> smokeping_prober? >> >> If not has anyone a working, simple configuration with unbound for this >> specific scenario? >> >> Is blackbox and smokeping using glibc methods to resolve DNS or something >> else? >> >> Thank you very much! >> >> -- >> 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-use...@googlegroups.com. >> To view this discussion on the web visit >> https://groups.google.com/d/msgid/prometheus-users/7972e866-f6be-47f6-8807-65f560f2aa3fn%40googlegroups.com >> >> <https://groups.google.com/d/msgid/prometheus-users/7972e866-f6be-47f6-8807-65f560f2aa3fn%40googlegroups.com?utm_medium=email&utm_source=footer> >> . >> > -- 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/48a894c9-be27-4dfc-8b27-a01cd3b9aef3n%40googlegroups.com.
