Often the bug would occur on only one of the twenty load-balanced servers, so just trying to reproduce the bug was hugely time consuming. They couldn’t scale. Customers were leaving. Morale was low and the best people on the teams were quitting. “ - Matt Jaynes - Taste Test: Puppet, Chef, Ansible, Salt
loses important configuration that somebody previously inserted in an ad hoc way Ops is not stable if different servers are maintained inconsistently Stable
this shouldn’t require much work Doing this on 5 machines shouldn’t require much work The risk should be higher waiting to do these upgrades than doing them regularly
Days http://devopsdays.org London Bangalore Downunder Mountainview Amsterdam Berlin Austin Paris London New Zealand Rome Mountain View Gurgaon Manila Goteborg Melbourne Boston Sao Paulo Hamburg Sydney Ghent