On Wed, 3 Mar 2021 at 14:41, Stuart Clark <stuart.cl...@jahingo.com> wrote:
> On 03/03/2021 14:33, John Dexter wrote: > > > > On Wed, 3 Mar 2021 at 14:15, Stuart Clark <stuart.cl...@jahingo.com> > wrote: > >> On 03/03/2021 13:00, John Dexter wrote: >> > While I understand the technical definitions I would appreciate some >> > real-world advice in the best way to structure my configuration file >> > for scalability and maintenance. >> > >> > Our system includes many instrumented modules running on an >> > application server on distinct ports, e.g on APPSERVER001 modules [A, >> > B, C, D, E, ..., Z] expose endpoints >> > APPSERVER001:[1001,1002,1003,1004, 1005, ..., 1026]/metrics >> > >> > However, we deploy multiple systems so we actually have APPSERVER001, >> > 002, ... each with the same modules exposing the same endpoints. I >> > might want a dashboard (Grafana) for each system, but I might also >> > want a dashboard displaying data across systems (e.g. show status of >> > module B on all systems) >> > Initially we only want to monitor one system as we get this all up and >> > running, before adding other systems over time. >> > Can anyone suggest how they would approach this in terms of job/scrape >> > config? file-based discovery seems a great option but I'm struggling >> > how to avoid a lot of copy-paste and duplicate information. >> > >> How are your applications deployed and managed? Kubernetes, Docker, >> Ansible, Puppet, etc.? >> >> -- >> Stuart Clark >> >> > Stuart, this is totally bespoke C++ Windows system, none of those :) The > file-discovery seems able to pull out my targets but I can't find a good > example how to structure my files. I suppose each Module A, B, C, ... > should be the jobs, but I'd far rather not have to list each of my > targets/servers once for every job when it's all the same. > > Yes each module probably wants to be a separate job. > > Multiple jobs can use the same YAML/JSON files with file_sd, so you could > create a single file (with just the hostnames/IPs) and then use relabling > to add the correct port number for each job. > > -- > Stuart Clark > > I've been looking at the examples here https://prometheus.io/docs/guides/file-sd/ and I cannot see how I would apply relabeling as described - their final example is similar to my case and they duplicate the entire section for the same job on two targets. You seem to be suggesting that could be avoided but I cannot see how. -- 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/CAGJRani2j-Px8xQbiWOsngNjbx2ZJqXTVFKDS0LuqSa_q-Ke%3Dg%40mail.gmail.com.
