Red Hat Enterprise 6 (aka Santiago) was released Nov. 10, 2010 —
3.5 years after the first release of RHEL 5 and 7 months after
RHEL 5.5. It will be supported to at least some extent for another 10 years
(until Nov. 30, 2020), which is a pretty impressive life cycle, especially compared to the 5-year support of Ubuntu’s LTS
releases. It’s just one confirmation of the intended market for RHEL6, which is clear throughout the release specs.
The latest version of the popular open source server OS was designed with enterprises
in mind. With a strong focus on scalability and flexibility, as well as support
for physical and virtual servers and cloud computing, there’s something for every
organization to get excited about.
Some of the basics: The supplied kernel is 2.6.32, and the new default
filesystem is ext4 (much faster than ext3, and scales to 16TB). Other
filesystem add-ons available include XFS (scalable to 100TB) and the clustered
GFS2 filesystem. NFSv4 and Fuse are also included.
Red Hat’s big focus in this release has been on scalability and flexibility,
supporting physical, virtual and cloud systems. There’s a stack of cluster
support options, based around the Corosync Cluster Engine, and including Conga
(Red Hat’s cluster management software), and virtualization options. The
kernel’s been improved to spend more time idle, which reduces power
usage — great for the power bills of big organizations, not to mention
the planet. RHEL6 can support up to 64 virtualized CPUs per guest using KVM
(which comes with the release; Xen doesn’t); and on hardware, up to 4,096
CPUs.
Red Hat is clearly thinking about what might happen during that
theoretical 10-year life cycle.
RHEL Installation and Setup
Download and install wasn’t as straightforward as I would have expected.
First, getting hold of a trial subscription was quite a hassle, and strange errors
showed up when registering. Also you have to tick a box saying that
you’ll allow business partners to contact you, which irritated me.
Note that you must download the full DVD (2.9GB) to install —
there is no network install option. There is a net-install image, but
this must be pointed at the full DVD image either on the system you’re
installing, or elsewhere on your local network. This is one of those moments
when the business user focus really shows. Install options include
basic/web/database server or a development workstation install.
I initially had disk problems with my first attempted install, and the
graphical installer gave either no error (first try) or an uninformative one
(second try). However, helpfully, Red Hat does provide a login prompt
and debugging information on the other virtual terminals (hit Ctrl-Alt-F2 for
the login prompt, and Ctrl-Alt-F3/F4/etc for debugging information), so I was
able to identify the problem. A basic server install took up around 1.2GB on
my system, and (correctly, from a security perspective!) didn’t provide much
in the way of specific server packages. A web or database server install
would presumably get those packages up and running.
I was, however, extremely surprised to find my Internet connection
wasn’t brought up by default. According to the official docs, manual
network configuration is often needed. Further, the docs describe the
graphical version of the Network Manager, rather than the console version of
system-network-config that you get with a server install. I
eventually got Ethernet up and running after manually providing a DNS server
address, which seemed a bit ridiculous. I can’t remember the last time I had
to fix Ethernet after a packaged Linux install. Similarly, there’s no
automatic system registration. Once the network’s running, you must run
rhn_register (whose error messages are not helpful) before you can
access the Yum repositories.