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?