Our goal to reduce the zero results rate was predicated on the understand that all of the traffic is coming from users. As shown in T110618, it's really not. There's a lot of automata hitting our API. In the report produced for the prior task, we differentiated between zero result queries coming from automata and from users. Let's add that information to our dashboard in the same fashion.
Description
Description
| Status | Subtype | Assigned | Task | ||
|---|---|---|---|---|---|
| Resolved | None | T106257 Send raw server side events to Kafka using a PHP Kafka Client {oryx} | |||
| Declined | None | T112846 Display automata and humans separately on zero results rate graph | |||
| Resolved | EBernhardson | T103505 Create analytics-centric Cirrus logs and have them import into HDFS | |||
| Resolved | EBernhardson | T106256 Kafka Client for MediaWiki | |||
| Resolved | • csteipp | T109384 Security review of apache/avro and nmred/kafka-php | |||
| Resolved | • bd808 | T111851 Package the Avro PHP library for easier Composer usage | |||
| Resolved | Ironholds | T110618 Make sense of why the zero results rate is still going up in spite of us having tackled prominent zero results generators | |||
| Resolved | Ironholds | T112295 Design and agree on an Avro schema for cirrus search request logging to hadoop | |||
| Resolved | • Nuria | T113521 Setup pipeline for search logs to travel through kafka and camus into hadoop {hawk} [55 pts] | |||
| Resolved | EBernhardson | T115715 Update CirrusSearchRequestSet schema to have a timestamp field |
Event Timeline
Comment Actions
Additional infrastructure is needed to do this. That is documented in T103505 and associated tasks.
Comment Actions
Pulling this out of the sprint because it's not possible until the infrastructure exists.
Comment Actions
Per the notice on https://discovery.wmflabs.org/ these dashboards are now in maintenance mode. This feature request is no longer relevant.