Currently, it's incredibly painful to debug Hadoop map/reduce jobs. The logfiles are on all the datanodes and failures can and do happen all over the place. This leads to a significant waste of time. Exposing Hadoop logfiles in logstash seems like a major win!
Version: wmf-deployment
Severity: enhancement
I think this is done now (w00t!) by assigning to @Gage for verification and proper attribution.
Yes, this project is considered complete. However because there are so many log sources, please do create tasks if you identify any event streams which are absent from Logstash.
https://wikitech.wikimedia.org/wiki/Analytics/Cluster/Logstash