For some reason, the type:ores keyword used in production doesn't work in the beta cluster logstash. It's annoying to workaround each time, we need an ORES dashboard for beta.
Description
Description
Event Timeline
Comment Actions
Currently, apache loglines are tagged with type "ores", e.g.
[pid: 10481] 10.68.21.68 (-) {34 vars in 661 bytes} [Mon Dec 4 16:50:24 2017] GET /v3/scores/enwiki/123456?features= => generated 16482 bytes in 7937 msecs (HTTP/1.1 200) 6 headers in 210 bytes (1 switches on core 0) user agent "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_13_1) AppleWebKit/604.3.5 (KHTML, like Gecko) Version/11.0.1 Safari/604.3.5"
The ORES celery logging is tagged with type "syslog", which is odd because we're only logging to the /srv/log/ores/app.log file. Example line:
[2017-12-04 16:50:24,975: INFO/MainProcess] Received task: ores.scoring_systems.celery_queue._lookup_score_in_map[enwiki:draftquality:0.1.0:123456] expires:[2017-12-04 16:50:39.952365+00:00]
Comment Actions
It needs refinement, but here's a rough pass which just shows all logs from the ores-beta host machine:
https://logstash-beta.wmflabs.org/app/kibana#/dashboard/2cc6f620-d9c9-11e7-b35f-4d258830b37d