My current environment primarily consists of two machines; a desktop and server both running Debian Wheezy. This has presented numerous problems when trying to access new versions of software. Often the versions I need or would like to use are not available in Wheezy’s repositories. I have enabled the testing and unstable repositories and defined the preferences to prevent all packages from being upgraded, but this has caused some annoying problems.
My goal is to maintain a stable system for personal use, but also have access to newer packages primarily for development. However, I have also found that packages for personal use were not available in stable.
tmux using SSH. However that interrupts my workflow because it duplicates a lot of the functionality of the parent window manager.rsyslog, a database). Although note that the filesystem provides excellent persistent storage (see Managing Data in Containers).The desktop will run Debian’s testing distribution, while the server will continue to run a stable distribution of Debian.
The alternative of installing CoreOS’s stable channel may present several problems. Every service should be, at least theoretically, in a separate container (e.g. sshd, git, a web server). CoreOS appears to support various levels of RAID through btrfs on the root filesystem, which excludes /usr and the ESP (see Working with btrfs and Common Troubleshooting, GitHub coreos/docs issue #222). However note that since CoreOS is designed to run on numerous distributed hosts it may be able to provide redundancy at a much higher level. Finally, running CoreOS on a single host means a loss of some its key functionality surrounding the use of a cluster.
Why should only one process run within a container? How do the distribution based containers run?