[08:43:18] started tomorrow's pad https://etherpad.wikimedia.org/p/SRE-Foundations-2019-07-31 [16:52:56] hey folks [16:53:05] where are we with the logstash postmortem? [16:53:10] I see nothing on https://wikitech.wikimedia.org/wiki/Incident_documentation still? [16:53:57] https://wikitech.wikimedia.org/wiki/Incident_documentation/20190723-logstash [16:54:22] oh I see it now [16:54:22] doh [20:14:13] godog: https://github.com/Selfnet/prometheus-network-exporter [20:14:39] :O [20:17:13] not sure how efficient it is, as it uses netconf over ssh [20:17:28] but it's better than snmp [20:17:30] lol, 'utitlities.py' [20:17:59] but not as good a the junos streaming telemetry stuff [20:20:54] where to start looking when promehtheus doesn't want to show me data? https://grafana.wikimedia.org/d/dxbfeGDZk/internal-anycast?orgId=1 [20:21:23] nothing in the logs, the prometheus hosts can query the exporter, the dashboard queries look good [20:24:44] ✔️ root@prometheus1003.eqiad.wmnet ~ 🕟⁉️ cat /srv/prometheus/ops/targets/bird_eqiad.yaml [20:24:47] # This file is managed by puppet [20:24:49] ✔️ root@prometheus1003.eqiad.wmnet ~ 🕟⁉️ [20:25:02] so the file is being generated empty, which seems wrong [20:28:53] yeah for sure [20:30:10] maybe because of that line https://github.com/wikimedia/puppet/blob/production/modules/profile/manifests/prometheus/ops.pp#L646 [20:30:38] does anything in the fleet actually have profile::bird? [20:31:27] nop [20:31:30] only profile::bird::anycast [20:31:40] cdanis: thanks! [20:33:01] one of these days i will understand how to use puppetboard's query page [20:34:37] cdanis: https://gerrit.wikimedia.org/r/c/operations/puppet/+/526521 [20:42:17] it works! [20:42:47] \o/ [21:50:54] cdanis: $ sudo cumin 'P:bird' [21:50:55] No hosts found that matches the query [21:50:56] :-P [21:54:23] jokes apart, the resources endpoint is not enabled in config, I'm trying to recall why. We disabled a bunch of them because potentially could expose private data [21:55:33] and resources is one of them IIRC, and that's the one you need to do that kind of query [21:55:48] (is the same endpoint that cumin uses fwiw)