linux.conf.au 2014

Perth, Western Australia - 6th to 10th January 2014

linux.conf.au


<-- Back to schedule

Processing Continuous Integration Log Events for Great Good

Project: OpenStack Infra

The OpenStack Continuous Integration systems run several thousand tests during the course of a normal day capturing more than a billion log events per week. This is a giant treasure trove of information that can give us CI system performance data, failure frequency, and detect bugs if the logs are stored in a queryable manner. Using Gearman, Logstash, and ElasticSearch as building blocks we have built a searchable index of our test log data. This index provides OpenStack developers with quick access to why their patches failed to pass tests, trend analysis on infrequent non deterministic failures, it can flag anomalous log events, and much much more. Suddenly the mountains of test logs are useful again.

This talk will describe the log processing architecture, what we did to make it handle over a billion events a week in near real time, and the neat things we have built around it (including automatic bug house keeping, trend analysis, and anomaly detection with CRM114).

Clark Boylan

Clark helps run the OpenStack developer automation providing robust continuous integration and development tools to a community of more than 800 developers. When not behind a keyboard he enjoys hiking and brewing beer.