I’m beginning to explore the logging options for docker and logging drivers seems like a sweet feature but I believe that there are some weak points that may be a stopper for me. I’m looking how to integrate splunk and elk with mesos and openshift. Maybe someone could point me if I’m wrong:
You loose the ‘docker logs’ command and the ability to check what is happening in a quickier manner from the console. I understand that you may want to check your logs in the logging plattform but some old school operation teams prefer to have their hands on console first
If the logging destination is temporaly unavailable your containers will NOT run. They would through and error. And what is worse, if your logging fails the running containers will not reconnect and you may loose some valuable logs.